Menu
How Aztec protects private data from quantum attacks today, and our plan for full post-quantum security.
All blockchains today – L1s, L2s, and shielded pools – make use of elliptic curves somewhere in their stack.
Elliptic curves are used in all sorts of places: signing transactions, validator attestations, data availability (such as blobs), zk-snarks, secret sharing, encryption, etc.
No cryptocurrency project today is 100% quantum-proof.
Nobody has a reliable date for when a quantum computer will break the elliptic-curve cryptography we use today, but we know it’s coming.
For a privacy network, some of the work cannot wait for that uncertainty to resolve. An attacker could collect encrypted data today and keep it until they have the means to decrypt it. This is the so-called “Harvest now; decrypt later” attack. Replacing an encryption scheme ten years from now would not protect the data an attacker already harvested under a classical encryption scheme.
Aztec is already quantum resistant against this type of attack. It was designed from the ground up to allow users to bunker-proof their accounts and data against quantum computers.
However, there is work to be done to make Aztec fully quantum secure and able to function with full privacy in the post-quantum world.
Our roadmap to expand our quantum security has two targets:
The first requires changes to how applications handle private data. The second reaches into the proving system, account authentication, key derivation, validator signatures and Ethereum settlement. Understanding why these need different schedules requires getting into the cryptography.
Years of hard-won experience have taught us that the greatest risk to a network like Aztec is implementation bugs. Our AI audit harness has found critical vulnerabilities, including ones that months of expert human internal and external audits didn’t find. These have been patched in the upcoming v6 Aztec Network release. We have spent the past 18 months hardening our software and cryptography. That work has to continue while we develop their replacements.
AI also changes cryptanalysis. For example, the attack on HAWK (a post-quantum signature scheme) exploited structure specific to that signature scheme and reduced its security enough to end its candidacy for standardization. That is a reason to examine the assumptions and structure of each scheme with care.
Rushing headfirst into new cryptography that has not faced enough scrutiny would expose users to immediate risks, greater than the speculative risks of quantum computers. The Foundation has finite resources, and time spent on the post-quantum migration is time we have to balance against securing the software users depend on today.
Our default approach to changing cryptography is conservative. Any changes to our protocols must be justified, subjected to extraordinary cryptanalysis and have formally-verified implementations.
To make Aztec post-quantum requires several pieces of cryptography. Aztec has an application layer (smart contracts), the Aztec protocol layer, and it settles on Ethereum itself. All three have responsibilities with respect to post-quantum security.
1. Proving System
2. Addresses & Keys
The latter application layer requirements are not on Ethereum’s post-quantum roadmap as they’re a unique requirement of privacy protocols like Aztec and ZCash.
Notice that the bulk of the changes required for user privacy are dictated by the choices of Aztec smart contract developers at the application layer. Apps can adopt post-quantum encryption and signature schemes today on Aztec, and different Aztec smart contract developers can freely explore differing approaches as the state of the art in post-quantum advances.
If and when quantum computers arrive, Aztec is already (uniquely for privacy protocols) protected in two key categories because of our existing design choices.
If I send you private tokens, there’s no onchain public balance update for you to see or query. Private state is stored in notes, so I need to tell you about the existence and contents of your new note for you to be able to use it.
A common solution is for me to publish an encrypted message to a public bulletin board that your wallet can find and decrypt. ZCash pioneered this approach.
Elliptic curve key exchange protocols are well-suited to this, because the amount of data I need to publish to convey “this is for you” and “here’s the ephemeral key you’ll need to decrypt this message” is a single ECDH ephemeral public key; an elliptic curve point; a tiny 32 bytes of info. We like this small communication overhead, because less data means faster tx throughput and cheaper DA costs passed on to users.
But elliptic-curve key exchanges can create a harvest-now-decrypt-later risk. The discrete log of such an ephemeral public key will one day be solvable by a quantum adversary. They could derive all ephemeral secret keys of all messages in the bulletin board. They’d then be able to brute-force combine the resulting list of ephemeral secret keys with a list of all known users’ viewing public keys to effectively derive all decryption keys on the network, to then decrypt all encrypted messages. They’d learn the contents of the note I sent you: the amount and that it belongs to you.
There are a few classical (pre-quantum) solutions to this. They follow from the observation that if an attacker doesn’t know one of: your public key or the ephemeral public key, then they won’t have enough information to decrypt any of your encrypted messages! Wallets are beginning to adopt the following:
Intentionally, most of the public keys in the Aztec protocol are hashed to reduce their leakage to a future post-quantum adversary. The public keys are not even exposed to Aztec smart contracts (which could be malicious).
(Private smart contract protocols even have a niche “harvest now; link later” attack to protect against, whereby a quantum adversary could link all your notes and nullifiers together to reconstruct your entire tx graph. Hashing the nullifier public key resolves this).
The Aztec Protocol is agnostic to the way in which users and applications (smart contracts) share private information between each other. It’s a beautifully flexible protocol because we knew times would change. Smart contract developers can adopt the above conventions today, and indeed many are doing so.
Aztec smart contracts can emit arbitrary bytes of data as “private logs”. These private logs are the intended delivery mechanism for sending encrypted notes. It’s up to smart contract developers how they populate those logs. I.e. they can choose:
Aztec also supports offchain note delivery, which gives applications another secure option. A sender delivers the message through an off-chain side channel that is not visible to harvest-now-decrypt-later attackers.
All this flexibility in today’s Aztec Protocol enables Aztec smart contract developers to explore the tradeoff space in a way that best serves their users and use cases. Choices will need to balance: proving speed; DA costs; discovery of pertinent notes; post-quantum threats.
Of course, most smart contract developers don’t want to think about such things, which is why the Aztec ecosystem has smart contract libraries with off-the-shelf solutions.
Our early 2027 work will focus on hardening these libraries and giving Aztec developers better patterns for protecting their users’ private data, with clear requirements for delivery and secret establishment.
A more comforting solution to post-quantum worries would be for Aztec smart contracts to instead adopt post-quantum secret sharing and encryption schemes.
We’re researching the best approaches and working on some tidy libraries. The good thing is, the agnosticism of the Aztec Protocol means multiple libraries for differing post-quantum secret sharing schemes can be written and adopted by different applications on Aztec.
In most other privacy protocols, the construction of an encrypted note is not constrained by any circuit.
For example, if I send you ZCash tokens, the ‘transfer’ function does not constrain that I have correctly encrypted your note, nor that I have attached the correct ephemeral public key that would enable you to discover and decrypt your encrypted note. I could technically broadcast garbage and you wouldn’t receive your funds. That’s arguably fine in a “simple” transfer protocol. It’s not in my interest to corrupt the encrypted notes that I create for you. After all, you can just claim I didn’t pay you until you do receive a valid and decryptable message.
But when you generalise beyond simple transfers to private smart contracts, there are use cases whereby a sender might not be incentivised to deliver valid data to a recipient. The sender might actually prefer to corrupt a recipient’s data. Imagine a private swap. The contract should not complete the trade if one party receives a commitment to a note whose contents they cannot recover.
In such cases, the private functions (circuits) need to constrain that a recipient will be capable of discovering and decrypting a sender’s encrypted messages. Aztec’s smart contract libraries have rails for this so-called “constrained delivery”. If a use case assumes distrust between sender and recipient, the smart contract developer can enforce their correct adherence to the rules. The protocol then constrains that all private logs emitted by smart contracts are guaranteed to be broadcast – without tampering – to Ethereum blobs.
With this in mind, the constraints required to support post-quantum encryption then become a very important consideration. We’ll need post-quantum libraries that are efficient to prove.
Key and ciphertext sizes matter, but so does the cost of proving the key exchange, encryption and associated checks. A scheme can be fast as native software and expensive inside a circuit.
MIKE, published on 6 October, illustrates why this remains an active research question. It is an isogeny-based non-interactive key-exchange proposal. Its authors report 80-byte public keys and shared-secret computation, including key validation, in 4.9 milliseconds on a 3.3 GHz AMD CPU at their claimed NIST level 1.
Compact keys are attractive for note delivery. We still need to understand its circuit costs and give its security assumptions time to face scrutiny. Those native timings cannot tell us whether it is the right choice for Aztec.
The encryption and proving-system decisions therefore affect each other. Choosing encryption first and discovering that it is prohibitively expensive in the chosen prover would leave us rebuilding one of them.
Choosing the encryption scheme we will prove inside future post-quantum circuits is a separate, later milestone. Applications that need the contract itself to guarantee delivery impose harder requirements on that choice.
Ok, so Aztec provides flexibility for good post-quantum privacy today. And we have plans to roll out even better post-quantum libraries.
But what about the proving system? A post-quantum adversary might be able to forge proofs. Thankfully we have some more time to choose a post-quantum proving system.
Aztec has several proving system requirements that other projects aren’t encumbered by (due to our commitment to both privacy, decentralisation and programmability), which we’ll elaborate on.
Common benchmarks for proving system performance don’t take into consideration all these requirements. Fast client-side recursion in particular is often overlooked.
Aztec uses CHONK, an elliptic-curve and pairing-based proving system built around the requirements of private smart contracts. Users generate proofs on their own devices, and a transaction can involve several private contract calls. Recursion lets us combine those calls into a proof of the transaction without requiring the browser to handle the entire computation as one enormous circuit. CHONK’s design puts low memory use and recursive composition at its centre.
There are tradeoffs you can make to get a faster prover. Some give the verifier more work. A custom “precompile” circuit for an expensive operation can save the prover from executing that operation through thousands of VM instructions, but introduces additional commitments and checks that increases Verifier complexity.
The problem is that, in the recursive setting, your verifier is in your prover. The next proof has to establish that the verifier accepted the previous proof.
I have lost count of the conversations about new prover technology where recursion gets a handwave and a “we’ll deal with it later.” It is a complex, messy problem that is easy to postpone until you need users to generate private transaction proofs in a browser.
Some alternative designs might execute an entire function stack within a single VM instance. That might work for some bounded stack depth. But snarks are gaining traction in web2 too (see zkPassport for example). Bridging between chains also often requires snarks. Privacy protocols should be able to verify traditional snarks for as long as we remain in a pre-quantum world.
Zero knowledge introduces another set of costs. I have heard the same assurances that adding it will be easy and will not affect performance. From personal experience, such claims often disintegrate on contact with reality.
We thought adding zero knowledge to CHONK would be easy. It took us six months and a re-architecture of our recursive proving stack. Modern proof systems have such complexity that throwing randomness into the execution trace and calling it a day almost never works. Adding ZK typically increases proof size and verifier work, which then increases the cost of recursion.
The result is that the real cost of recursion in a production proving system, with all of the edge-cases and exceptions and little additions that pile up, is never as fast as the initial proof-of-concept claims. For Aztec that is a problem because recursion is essential and not an afterthought - we won’t commit to a proving system that can’t meet our extremely tight recursion requirements.
Luckily there are some candidates, and the Ethereum Foundation is doing trailblazing work as one of the few teams that has started to take this problem seriously.
Proof size matters because Aztec transactions travel through a peer-to-peer network. Every node that receives and forwards a proof pays its bandwidth cost. Aztec’s 20kb proofs are “ok”. If it ever becomes a bottleneck we can make them smaller with little effort because we use elliptic curves.
Post-quantum proof systems (the fast ones), on the other hand, have proofs of ~300KB, which is not ok.
Worse, there are zero post-quantum proving systems whose proofs are small enough to gossip across a p2p network whilst achieving acceptable transaction throughput. There are fundamental constraints which mean this is not likely to change!
Fast recursion offers another path. A node could receive two transaction proofs, produce one proof that establishes the validity of both, and forward the aggregate. Other nodes could repeat the process. Transaction data would still have to travel through the network, but the proof overhead could shrink across a batch.
Ethereum’s post-quantum research faces a related problem. Its proposed hash-based validator signatures lack BLS’s cheap native aggregation, so leanVM supplies a proof that many signatures pass verification. The draft EIP-8292 assigns that proving work to aggregators with suitable hardware. Their longer-term roadmap requires full virtual-machine proof aggregations which presents an extremely similar problem to ours.
For the Aztec approach to work on ordinary nodes, aggregation needs to be cheap. My working estimate is that two-to-one aggregation in roughly 20–100 milliseconds on a modern CPU could make it practical.
Nodes receive different transactions at different times, so they will duplicate work. An attacker may try to exploit that duplication. We need to establish how many aggregations one malicious transaction can induce and what happens to throughput under that load.
This problem is currently under-explored and poorly parametrised and substantial work needs to occur before a definitive solution can be decided on.
The main candidates are hash-based systems over binary fields and small prime fields, lattice-based systems, and proposals that combine both prime-field and binary-field representations.
The Flock paper gives a good example of the progress in binary-field proving. Its authors report proving 660,000 BLAKE3 hashes per second on an M4 mac on ten cores.
These are impressive numbers. Hashing is a major part of verifying hash-based proofs, so cheap hashing is good for recursion.
However, binary field systems are weak with regards to integer multiplication. Trustless access to Web2 data is one of the biggest untapped capabilities of blockchain. Private applications should be able to use passport signatures, credit credentials, TLS-based data proofs and signed web tokens without publishing the underlying information. That means dealing with ugly Web2 cryptography: RSA, large-field elliptic curves and integer arithmetic.
Binary-field arithmetic does not behave like integer arithmetic. In particular, integer addition has carries, and multiplication needs a representation and a circuit that handles them. A BLAKE3 benchmark tells us little about the cost of proving a 4096-bit RSA verification. We need to measure the operations our applications will use.
That is what makes hybrid approaches interesting to me: using one commitment both in a binary representation and a prime field representation depending on context. This is an active area of research with limited existing published work, which highlights how rapidly evolving this space is.
Lattice-based proving deserves serious attention too. The Jolt team’s September release replaced the elliptic-curve commitment scheme Dory with Akita, based on Module-SIS. The authors reported faster proving and verification and proofs below 100 KB. Their announcement also described zero knowledge as forthcoming, so those figures do not settle our comparison.
Lattice commitments retain algebraic structure that could be useful for the kinds of composition we rely on. That makes them attractive candidates to investigate, although translating CHONK’s techniques would require work. The field size is a design choice we need to evaluate within each construction; lattices do not impose one universal 128-bit-field requirement.
Much of the suspicion around lattice systems rests on vibes about whether the underlying structure is vulnerable to AI cryptoanalysis. There are good reasons to scrutinize a particular assumption, parameter set or reduction. “It uses lattices” does not supply that scrutiny.
Not all lattice crypto is equal. The fundamental short-integer-solution hardness assumption has had decades of cryptanalysis. Lattice crypto has a degraded reputation because some modern protocols layer on additional, weaker hardness assumptions that have shown to be broken by quantum adversaries. But there is no quantitative evidence to suspect the core harness assumptions of lattice crypto are vulnerable.
Either way, the rate of development gives me confidence that we will have good options. It also makes choosing a winner today premature. We need comparisons that include our applications, recursion, zero knowledge and security parameters together.
In Aztec transactions, digital signatures are private witnesses within private functions of smart contracts (zero-knowledge circuits), which helpfully means we don’t care so much about the size of an individual signature. This gives us more freedom than non-rollups e.g. Ethereum, where signature size directly contributes to their transaction size. This matters because post-quantum digital signatures are typically large.
Similar to other privacy protocols such as ZCash Sapling and Orchard, several Aztec keys are baked-into the preimage of an Aztec address. To enable ‘post-quantumification’ of the Aztec protocol, we’re exploring externalising those keys to a canonical keystore registry. An Aztec address would then cease to be an elliptic-curve public key, and instead would simply be a hash which serves as a lookup into this registry. The registry could store all manner of exotic post-quantum public keys for different use cases.
Aztec uses Poseidon2 throughout the protocol. Poseidon2 is technically post-quantum secure. And while Poseidon2 is secure for our chosen domain, recent breaks over small-field Poseidon signal that we should move to a different hash function in the long-term.
We’re confident that Poseidon2 will remain practically secure for many years, but the writing is on the wall and it would not be wise to use this hash function in Aztec’s future post-quantum system.
We will likely replace it with either Blake2s or Blake3. The Ethereum Foundation are also considering a canonical hash function for their post-quantum roadmap and we’ll more than likely follow their lead here. Binary-field proof systems can trivially handle BLAKE hashes so the need for Poseidon2 will have run its course.
Our proposed selection milestones are:
The encryption milestone needs circuit measurements before we settle the proving-system choice. The proving-system decision needs working recursion and zero knowledge, alongside representative application workloads. These dates give us a sequence for answering those questions.
After selection, we need to replace the cryptography across the system. That includes addresses and key derivation, application-specific key checks inside the kernels, account authentication, validator authentication and the proof chain that ends on Ethereum. We also need to handle existing private state and give users and applications a migration path.
Ethereum supplies another dependency. Aztec’s final proof verification and blob commitments sit within Ethereum’s cryptographic infrastructure. A full post-quantum migration has to account for Ethereum’s upgrades as well as our own.
The roughly three years between the final selection milestone and end of year 2030 leave room for integration to expose problems that isolated benchmarks cannot. Development and integration of multiple novel post-quantum cryptographic primitives into Aztec will be neither simple nor fast. In addition, the entire tech stack will have to be formally verified due to the incredible threat that AI poses with regards to finding and exploiting bugs.
I think Q-day within four years is unlikely, and the roadmap is designed under the assumption that post-quantum work is not the sole focus and obsession of the Aztec Foundation. If there is evidence of faster progress in quantum hardware that would bring forward Q-day, we will be able to accelerate our roadmap by diverting resources to this project.
Developers will not have to wait for that full migration to make better choices about private data. The January work will provide hardened libraries and documented patterns to build against. A developer building a private application will be able to protect the data they create today to stand the test of time.
Core contributors identified a critical vulnerability affecting the V5 Alpha proving system on 27 July 2026 through internal AI-assisted auditing.
V5 remains Alpha software. Critical findings can arise during this phase, and the audit process exists to identify them before broader deployment. This finding places V5 funds, applications, and contract state at risk.
Treat funds and applications on V5 as exposed to a protocol-level failure until contributors complete incident response work and operators carry out the required network actions.
An attacker may be able to exploit a flaw in the current V5 proving system by constructing a proof that passes verification for a transaction the network should reject. If accepted, that transaction could produce a state transition outside the rules V5 intends to enforce.
Contributors cannot determine whether anyone exploited the flaw before this finding. The affected system lacks the information needed to distinguish ordinary accepted transactions from transactions accepted through the flawed proving path. Historical chain activity cannot establish whether exploitation occurred or quantify its impact.
We expect application teams to prepare safeguards in the coming weeks.
Those safeguards may include changes to application controls, deployment procedures, user flows, and migration plans. We expect each team to assess its contracts and determine which protections fit its architecture and users.
We expect teams planning a V5 deployment to pause that work until contributors publish further guidance. We expect teams with live contracts to review their ability to limit user exposure, isolate affected functionality, and move users to fresh deployments if needed.
We expect applications that maintain administrative or emergency controls to assess whether those controls can reduce user risk during the incident timeframe.
Core contributors are working with operators, application teams, and bridge operators as applications add security guards around affected flows.
The findings from this incident will inform the V6 release, including circuit updates that prevent the network from accepting proofs tied to an affected proving system.
V5 launched as Alpha software, with V6 planned for later in 2026. Contributors will publish a security roadmap covering the remaining work and release path.
Reviewers have not identified other high-severity or critical V5 Alpha vulnerabilities at this time.
Internal and external human audits have completed, and contributors continue AI-assisted auditing. Alpha is the period for identifying faults before production deployment.
Dark Forest is a real-time strategy game played across a procedurally generated universe where most of the map is hidden. You cannot see rival players, their planets, or their fleets. You only know what you have explored. Everyone shares one universe, and nobody has the full picture.
In most onchain games, every position and every move is public, because the chain is public. Dark Forest used zero-knowledge proofs to break that assumption: players prove their moves are valid without revealing where those moves came from. The result is a game of hidden information running on a public network.
Dark Forest Aztec ports the original Dark Forest 0.6 to Aztec. It keeps the gameplay from the original and rebuilds the privacy layer on Aztec's programmable privacy.
A note before diving in: this is early, experimental software on Aztec Alpha V5. Treat it as an alpha and play accordingly.
You start on a single home planet with almost the entire map dark. To find anything you mine the universe, running a client that explores coordinates and reveals what sits there: unclaimed planets, resources, and eventually the edges of other players' territory.
You are never handed a view of the board. You earn it one region at a time, and everyone else works under the same fog.
Your home coordinates and your fleet movements are private state, expressed as first-class private notes on Aztec. Your location and where you send energy stay hidden, enforced in the contracts by zero-knowledge cryptography.
What sits onchain is a set of cryptographic commitments. Instead of storing every planet's full details in the open, the contracts store Poseidon2 hashes of entity state. When you make a move, your client supplies the full state, the contract checks it against the stored hash, applies the change under zero-knowledge constraints, and writes a new hash back. Full game state lives offchain and gets rebuilt from public logs by an indexer, which is what renders your map without exposing every player's position.
So you can prove you made a legal move from a planet you own without revealing where that planet is. Aztec applies the same principle to private payments and private contracts.
Four actions carry the game.
Explore. Your explorer sits in the bottom left. Set it running and it uncovers the map around you, surfacing planets, resources, and other players.
Send energy. Most planets produce energy. Click and drag from a planet you own toward a target to capture or weaken it.
Route silver. Asteroid fields produce silver. Move it to your planets and spend it on upgrades, or send it to a Spacetime Rip to convert it into score.
Hunt artifacts. Some planets hold artifacts. Your Gear ship discovers them. Once harvested, you deposit them on planets to boost stats.
Four stats drive most decisions.
Energy is the core resource. Planets generate it over time up to a capacity, and you spend it on everything: claiming planets, reinforcing your own, attacking rivals. Two details matter. Moves are taxed, so a flat percentage of a planet's total capacity burns every time you send energy, which discourages small frequent moves. And energy decays over distance, so send it too far and almost nothing arrives. A common rule of thumb is to let a planet fill to about 75%, then send it down to about 25%.
Defense reduces the damage incoming energy does when it lands. Higher-level planets often have lower defense, but they hold much more energy, so they still take more to capture. Defense matters most on front lines.
Range sets how far a planet can send energy. It governs how fast you expand and how efficiently you move energy inside your own empire, since shorter relative distances mean less decay. Good range also lets you strike deep into an opponent's territory.
Speed sets how quickly a move arrives. Usually secondary, though a fast strike can land before a rival reacts, and some playstyles reward capturing many nearby planets quickly.
Planets can also be upgraded with silver and enhanced with artifacts. Space types carry different multipliers, from mild Nebula to punishing Dead Space, so where a planet sits changes how it plays.
There is a scoreboard, and territory alone does not win it. This round scores two activities: discovering artifacts with your Gear ship, and withdrawing silver through Spacetime Rips.
Point values from the in-game help page:
Silver accrues one point at a time. A single Mythic artifact is worth twenty million of them, so artifact hunting decides rounds and silver withdrawal sets your floor.
Silver has two competing uses. Spend it on upgrades and your planets get stronger, extending range and hardening defense. Withdraw it through a Spacetime Rip and it becomes scored points, but it is gone. Every unit is a choice between building the empire and banking points.
Upgrades tend to win early, since a stronger empire reaches more asteroid fields and finds more artifacts. Late in a round that calculation flips, because a planet you never use is worth less than points already scored.
Artifacts do both jobs at once. They score on discovery, and once deposited they boost a planet's stats, which makes the next expedition easier.
Nothing happens until you find something to act on. Your explorer turns dark space into planets you can capture, asteroid fields you can mine, and artifact-bearing planets you can raid. Sitting still means no new energy, no silver, no score.
Exploring also buys information. The map you have uncovered is an advantage nobody else holds. Knowing where high-level planets sit, which asteroid fields are unclaimed, and where space types shift lets you plan further ahead than someone still working through their starting region.
You find other players as a byproduct. There is no player list. You explore outward until your revealed region touches territory someone already owns: a planet in another player's colors, sitting where you were about to expand. Their home coordinates stay private, so you learn something narrow. Someone is here, roughly this direction, holding this much. You infer the rest, and you have no way of knowing whether they found you first.
You have three broad options.
Stay quiet and keep growing. Nothing forces you to engage. Keep exploring elsewhere, keep routing silver, keep upgrading. Your positions stay private, so silence costs you only time, which is what you want if they are stronger. The risk is that they are doing the same thing faster.
Fortify the border. If the contact sits somewhere you cannot lose, spend energy hardening the planets facing them. Defense is worth most where an attack will actually land. This keeps the option to fight without committing to one.
Attack. Send enough energy to overwhelm the target's defense and the planet becomes yours, along with its production and its position as a staging post. Higher-level planets are the prize and take proportionally more to crack.
Attacking costs more than energy. A move that lands tells your rival where you strike from, and that you are close enough to be worth answering. Retaliation can then come from directions you have not explored, launched from planets you cannot see.
Everyone plays one shared universe in real time. No turns, no lobbies. Energy regenerates whether you are watching or not, moves stay in flight while you sleep, and rivals expand while you are away from the screen.
Most strategy games let you watch a threat approach. Here you tend to see the consequences: a planet you owned this morning in someone else's colors, an incoming move you notice once it is already close.
That produces a particular kind of paranoia. You are trying to find everyone else while avoiding being found, and every expansion is a strategic bet that the space ahead is empty.
Information becomes tradeable, because it is scarce. Players compare notes, warn each other about aggressive neighbors, and agree who expands where, then break those agreements when the scoreboard makes it worth breaking.
A fully onchain game where players cannot see each other's positions is hard to build, and building it well says something about the platform underneath.
Hidden state, private notes, and client-side proving are the same building blocks behind private applications across Aztec. Dark Forest is a way to watch them work.
Dark Forest Aztec is playable now as an alpha. Expect a learning curve; the original was famous for it. DFArchon maintains onboarding material and a community for new players. Round One is live. The universe is dark, and everyone else is out there somewhere. Go find them, quietly.
Gas on Aztec is known as Fee Juice and is used to pay for transaction costs. This is the same as $ETH on Ethereum. Some apps will handle transaction costs for you under the hood, but if you are using a browser extension wallet, you will not be able to send transactions without it. Fee Juice can be obtained by bridging the $AZTEC token on Ethereum to the Aztec Network L2. This means that under the hood, all activity that happens on Aztec is underpinned by the $AZTEC token bridged into the network. Some bridges like Shield (by human.tech) handle this for you, allowing you to allocate a portion of your bridged transaction to convert into Fee Juice and land in your wallet automatically.
Assets and transactions on the Aztec Network can be either public or private. If you bridge publicly, your tokens will arrive as public, traceable tokens visible to all. Privately bridging, on the other hand, will give you private assets that are visible only to you. These assets can then be sent privately to another user or wallet without ever revealing who you are, what tokens were sent, how many, or who the recipient is.
Like tokens on the Aztec Network, Fee Juice (gas) can also be public or private. The reason for this is that even if what you are sending is private, the gas you spend to execute that transaction could still be visible if you are using public Fee Juice, potentially revealing transaction patterns and activity. Private Fee Juice keeps your entire transaction footprint hidden. When you send a private transaction, you can use private Fee Juice, and when you send a public transaction, you can use public Fee Juice, which means your transaction costs are always aligned with the type of transaction you're making.
Aztec has native fee abstraction, which means apps could let you pay for transactions in any token you want, or cover your fees entirely. Apps like Nyx may choose to cover part or all of a user's transaction costs, or allow you to pay in tokens that are convenient for you. This means you will most likely never see Fee Juice in an app; instead, you'll pay in whatever makes sense for what you're doing, on your terms. Similarly, you might never even see an Aztec wallet at all, because the app itself becomes your interface that you connect to using your MetaMask wallet.
If you're using a browser extension like Azguard, you'll manage Fee Juice directly in your wallet alongside your private and public balances, converting between tokens as needed to cover transaction costs.
When you bridge tokens in, you'll need enough Fee Juice to cover the cost of your first transaction, then you'll need to monitor how much Fee Juice you have available to make transactions. Browser wallets will allow you to send either publicly or privately to other users and will default to using either public or private Fee Juice depending on the type of transaction. Both private Fee Juice and public Fee Juice will appear by default in your token list.
How you handle Fee Juice depends on where you're transacting: apps can abstract it away entirely and let you pay in any token, while a browser wallet like Azguard puts it in your hands to manage across public and private balances. Match your gas to your transaction, keep private activity private down to the fee, and you move on your terms.
The Aztec Network today activated Alpha V5, a major protocol upgrade passed by token-holder governance and executed onchain. Alpha V5 reduces private-transaction proving times by more than 2x compared to the previous version, lowers the cost of a fully private transaction by roughly 50%, resolves the critical issues found in V4, and sees the first wave of apps go live. Users can now send private transactions and earn yield on Aave simply by connecting their Ethereum wallets on Nyx, bridge from Ethereum to Aztec using Shield or TRAIN, privately collect NFTs on RavenHouse, or play Dark Forest Aztec, a hidden-information strategy game in a universe that lives entirely onchain.
"Alpha V5 continues Aztec's work at the frontier of client-side proving, with cryptographic breakthroughs that cut proving times by more than half this release," said Zac Williamson, Co-founder, Aztec Foundation. "We believe Aztec is now the fastest system in the world for proving a fully private transaction entirely on a user's own device, and every release moves the industry closer to private transactions at public transaction speeds."
As the only decentralized privacy L2, Aztec is the credibly neutral privacy layer for Ethereum. Aztec allows anyone to write smart contracts that include both private and public aspects – every private transaction is proven on the user's own device, so no operator, sequencer, or intermediary can see the data. The Alpha V5 proving improvements come from cryptographic advances that make this client-side proving faster than any prior release. The network remains in alpha, but with V5 it is ready for teams to begin building and deploying applications.
Making private transactions practical comes down to how quickly a proof can be generated on a user's own device, without offloading that work to a server that would learn what the user is doing. On Alpha V5, proving a private token transfer natively now takes approximately 2.5 seconds on a consumer laptop, down from 5.2 seconds on V4, and about 6.8 seconds in a browser, down from 12.5 seconds. Across every measured transaction flow, client-side proving times improved by approximately 2x compared with V4.
Alpha V5 lowers ECDSA signature-verification cost by approximately 2x, speeds up Poseidon2 hashing by approximately 3x, and reduces the protocol circuit gate count by approximately 50% (gate count is the number of individual operations a proving circuit must perform, and it is the main driver of how long a proof takes to generate). Each of these lowers the amount of work a device performs to prove a transaction, and the reduction in gate count in particular compounds across every proof the network generates.
Alpha V5 launches the first wave of apps on a network where privacy is built into the protocol rather than managed by an operator. On other networks that claim privacy, transactions still pass through an operator or node that reads them in plaintext, or depend on a viewing key that a third party holds, so users rely on someone else to protect their data and to decide when it gets disclosed. On Aztec, every private transaction is proven on the user's own device, so the app, the sequencer, and any operator never need to see the underlying data. Nyx is one of these apps, allowing users to privately send transactions and privately earn yield on Aave.
"On Ethereum, everything you do is public. That's why we built Nyx: a private account governed by your Ethereum wallet", said Nikhil, Co-founder of Nyx. "Now you can send, receive and earn in private. Nyx was the first app live on the Aztec Alpha, and we're excited to expand participation to more users with the added stability of Alpha V5."
Other apps on Alpha V5 include Azguard and Nethermind (wallets), Shield, TRAIN, and RavenHouse (bridges), and the Aztecscan block explorers. Also launching is Dark Forest Aztec, a game where users explore a universe, control planets, manage planetary energy, expand territory, and launch attacks through strategic play with private state and hidden actions.
Transaction fees on Aztec come from two main sources: the cost of proving a transaction and the cost of verifying the rollup proof on Ethereum. Alpha V5 reduces both. It lowers the network's proving-cost parameter by 50%, and it reduces the L1 gas required to verify a rollup proof by approximately 40%. Because rollup proofs are verified on Ethereum and that cost is shared across all transactions in a batch, the L1 reduction lowers fees for every user, while the lower proving-cost parameter reduces the per-transaction proving fee directly. Together, these bring the average cost of a fully private token transfer to under a $0.05 transaction cost.
Alpha V5 also hardens the network on several fronts. It resolves critical vulnerabilities found in Alpha V4 along with additional bugs discovered since launch. Aztec's bug bounty program on Cantina also drew more than 234 security researchers to participate. The network remains in alpha, and further bugs may surface as usage grows, but each release has closed the issues found in the last and strengthened the protocol against new ones. With the critical V4 issues resolved and these safeguards in place, Alpha V5 is stable enough for teams to begin building and deploying applications.
Alpha V5 is live now, view the Alpha V5 landing page for a full list of features, performance updates, and live apps to explore.
Aztec is the only decentralized, privacy-first Layer 2 on Ethereum. Developers write private and public logic in the same smart contract, and private functions are executed and proven on the user's own device, so no operator sees the underlying data. The protocol is upgraded through onchain governance, and the network settles to Ethereum. For more information, visit aztec.network.
All blockchains today – L1s, L2s, and shielded pools – make use of elliptic curves somewhere in their stack.
Elliptic curves are used in all sorts of places: signing transactions, validator attestations, data availability (such as blobs), zk-snarks, secret sharing, encryption, etc.
No cryptocurrency project today is 100% quantum-proof.
Nobody has a reliable date for when a quantum computer will break the elliptic-curve cryptography we use today, but we know it’s coming.
For a privacy network, some of the work cannot wait for that uncertainty to resolve. An attacker could collect encrypted data today and keep it until they have the means to decrypt it. This is the so-called “Harvest now; decrypt later” attack. Replacing an encryption scheme ten years from now would not protect the data an attacker already harvested under a classical encryption scheme.
Aztec is already quantum resistant against this type of attack. It was designed from the ground up to allow users to bunker-proof their accounts and data against quantum computers.
However, there is work to be done to make Aztec fully quantum secure and able to function with full privacy in the post-quantum world.
Our roadmap to expand our quantum security has two targets:
The first requires changes to how applications handle private data. The second reaches into the proving system, account authentication, key derivation, validator signatures and Ethereum settlement. Understanding why these need different schedules requires getting into the cryptography.
Years of hard-won experience have taught us that the greatest risk to a network like Aztec is implementation bugs. Our AI audit harness has found critical vulnerabilities, including ones that months of expert human internal and external audits didn’t find. These have been patched in the upcoming v6 Aztec Network release. We have spent the past 18 months hardening our software and cryptography. That work has to continue while we develop their replacements.
AI also changes cryptanalysis. For example, the attack on HAWK (a post-quantum signature scheme) exploited structure specific to that signature scheme and reduced its security enough to end its candidacy for standardization. That is a reason to examine the assumptions and structure of each scheme with care.
Rushing headfirst into new cryptography that has not faced enough scrutiny would expose users to immediate risks, greater than the speculative risks of quantum computers. The Foundation has finite resources, and time spent on the post-quantum migration is time we have to balance against securing the software users depend on today.
Our default approach to changing cryptography is conservative. Any changes to our protocols must be justified, subjected to extraordinary cryptanalysis and have formally-verified implementations.
To make Aztec post-quantum requires several pieces of cryptography. Aztec has an application layer (smart contracts), the Aztec protocol layer, and it settles on Ethereum itself. All three have responsibilities with respect to post-quantum security.
1. Proving System
2. Addresses & Keys
The latter application layer requirements are not on Ethereum’s post-quantum roadmap as they’re a unique requirement of privacy protocols like Aztec and ZCash.
Notice that the bulk of the changes required for user privacy are dictated by the choices of Aztec smart contract developers at the application layer. Apps can adopt post-quantum encryption and signature schemes today on Aztec, and different Aztec smart contract developers can freely explore differing approaches as the state of the art in post-quantum advances.
If and when quantum computers arrive, Aztec is already (uniquely for privacy protocols) protected in two key categories because of our existing design choices.
If I send you private tokens, there’s no onchain public balance update for you to see or query. Private state is stored in notes, so I need to tell you about the existence and contents of your new note for you to be able to use it.
A common solution is for me to publish an encrypted message to a public bulletin board that your wallet can find and decrypt. ZCash pioneered this approach.
Elliptic curve key exchange protocols are well-suited to this, because the amount of data I need to publish to convey “this is for you” and “here’s the ephemeral key you’ll need to decrypt this message” is a single ECDH ephemeral public key; an elliptic curve point; a tiny 32 bytes of info. We like this small communication overhead, because less data means faster tx throughput and cheaper DA costs passed on to users.
But elliptic-curve key exchanges can create a harvest-now-decrypt-later risk. The discrete log of such an ephemeral public key will one day be solvable by a quantum adversary. They could derive all ephemeral secret keys of all messages in the bulletin board. They’d then be able to brute-force combine the resulting list of ephemeral secret keys with a list of all known users’ viewing public keys to effectively derive all decryption keys on the network, to then decrypt all encrypted messages. They’d learn the contents of the note I sent you: the amount and that it belongs to you.
There are a few classical (pre-quantum) solutions to this. They follow from the observation that if an attacker doesn’t know one of: your public key or the ephemeral public key, then they won’t have enough information to decrypt any of your encrypted messages! Wallets are beginning to adopt the following:
Intentionally, most of the public keys in the Aztec protocol are hashed to reduce their leakage to a future post-quantum adversary. The public keys are not even exposed to Aztec smart contracts (which could be malicious).
(Private smart contract protocols even have a niche “harvest now; link later” attack to protect against, whereby a quantum adversary could link all your notes and nullifiers together to reconstruct your entire tx graph. Hashing the nullifier public key resolves this).
The Aztec Protocol is agnostic to the way in which users and applications (smart contracts) share private information between each other. It’s a beautifully flexible protocol because we knew times would change. Smart contract developers can adopt the above conventions today, and indeed many are doing so.
Aztec smart contracts can emit arbitrary bytes of data as “private logs”. These private logs are the intended delivery mechanism for sending encrypted notes. It’s up to smart contract developers how they populate those logs. I.e. they can choose:
Aztec also supports offchain note delivery, which gives applications another secure option. A sender delivers the message through an off-chain side channel that is not visible to harvest-now-decrypt-later attackers.
All this flexibility in today’s Aztec Protocol enables Aztec smart contract developers to explore the tradeoff space in a way that best serves their users and use cases. Choices will need to balance: proving speed; DA costs; discovery of pertinent notes; post-quantum threats.
Of course, most smart contract developers don’t want to think about such things, which is why the Aztec ecosystem has smart contract libraries with off-the-shelf solutions.
Our early 2027 work will focus on hardening these libraries and giving Aztec developers better patterns for protecting their users’ private data, with clear requirements for delivery and secret establishment.
A more comforting solution to post-quantum worries would be for Aztec smart contracts to instead adopt post-quantum secret sharing and encryption schemes.
We’re researching the best approaches and working on some tidy libraries. The good thing is, the agnosticism of the Aztec Protocol means multiple libraries for differing post-quantum secret sharing schemes can be written and adopted by different applications on Aztec.
In most other privacy protocols, the construction of an encrypted note is not constrained by any circuit.
For example, if I send you ZCash tokens, the ‘transfer’ function does not constrain that I have correctly encrypted your note, nor that I have attached the correct ephemeral public key that would enable you to discover and decrypt your encrypted note. I could technically broadcast garbage and you wouldn’t receive your funds. That’s arguably fine in a “simple” transfer protocol. It’s not in my interest to corrupt the encrypted notes that I create for you. After all, you can just claim I didn’t pay you until you do receive a valid and decryptable message.
But when you generalise beyond simple transfers to private smart contracts, there are use cases whereby a sender might not be incentivised to deliver valid data to a recipient. The sender might actually prefer to corrupt a recipient’s data. Imagine a private swap. The contract should not complete the trade if one party receives a commitment to a note whose contents they cannot recover.
In such cases, the private functions (circuits) need to constrain that a recipient will be capable of discovering and decrypting a sender’s encrypted messages. Aztec’s smart contract libraries have rails for this so-called “constrained delivery”. If a use case assumes distrust between sender and recipient, the smart contract developer can enforce their correct adherence to the rules. The protocol then constrains that all private logs emitted by smart contracts are guaranteed to be broadcast – without tampering – to Ethereum blobs.
With this in mind, the constraints required to support post-quantum encryption then become a very important consideration. We’ll need post-quantum libraries that are efficient to prove.
Key and ciphertext sizes matter, but so does the cost of proving the key exchange, encryption and associated checks. A scheme can be fast as native software and expensive inside a circuit.
MIKE, published on 6 October, illustrates why this remains an active research question. It is an isogeny-based non-interactive key-exchange proposal. Its authors report 80-byte public keys and shared-secret computation, including key validation, in 4.9 milliseconds on a 3.3 GHz AMD CPU at their claimed NIST level 1.
Compact keys are attractive for note delivery. We still need to understand its circuit costs and give its security assumptions time to face scrutiny. Those native timings cannot tell us whether it is the right choice for Aztec.
The encryption and proving-system decisions therefore affect each other. Choosing encryption first and discovering that it is prohibitively expensive in the chosen prover would leave us rebuilding one of them.
Choosing the encryption scheme we will prove inside future post-quantum circuits is a separate, later milestone. Applications that need the contract itself to guarantee delivery impose harder requirements on that choice.
Ok, so Aztec provides flexibility for good post-quantum privacy today. And we have plans to roll out even better post-quantum libraries.
But what about the proving system? A post-quantum adversary might be able to forge proofs. Thankfully we have some more time to choose a post-quantum proving system.
Aztec has several proving system requirements that other projects aren’t encumbered by (due to our commitment to both privacy, decentralisation and programmability), which we’ll elaborate on.
Common benchmarks for proving system performance don’t take into consideration all these requirements. Fast client-side recursion in particular is often overlooked.
Aztec uses CHONK, an elliptic-curve and pairing-based proving system built around the requirements of private smart contracts. Users generate proofs on their own devices, and a transaction can involve several private contract calls. Recursion lets us combine those calls into a proof of the transaction without requiring the browser to handle the entire computation as one enormous circuit. CHONK’s design puts low memory use and recursive composition at its centre.
There are tradeoffs you can make to get a faster prover. Some give the verifier more work. A custom “precompile” circuit for an expensive operation can save the prover from executing that operation through thousands of VM instructions, but introduces additional commitments and checks that increases Verifier complexity.
The problem is that, in the recursive setting, your verifier is in your prover. The next proof has to establish that the verifier accepted the previous proof.
I have lost count of the conversations about new prover technology where recursion gets a handwave and a “we’ll deal with it later.” It is a complex, messy problem that is easy to postpone until you need users to generate private transaction proofs in a browser.
Some alternative designs might execute an entire function stack within a single VM instance. That might work for some bounded stack depth. But snarks are gaining traction in web2 too (see zkPassport for example). Bridging between chains also often requires snarks. Privacy protocols should be able to verify traditional snarks for as long as we remain in a pre-quantum world.
Zero knowledge introduces another set of costs. I have heard the same assurances that adding it will be easy and will not affect performance. From personal experience, such claims often disintegrate on contact with reality.
We thought adding zero knowledge to CHONK would be easy. It took us six months and a re-architecture of our recursive proving stack. Modern proof systems have such complexity that throwing randomness into the execution trace and calling it a day almost never works. Adding ZK typically increases proof size and verifier work, which then increases the cost of recursion.
The result is that the real cost of recursion in a production proving system, with all of the edge-cases and exceptions and little additions that pile up, is never as fast as the initial proof-of-concept claims. For Aztec that is a problem because recursion is essential and not an afterthought - we won’t commit to a proving system that can’t meet our extremely tight recursion requirements.
Luckily there are some candidates, and the Ethereum Foundation is doing trailblazing work as one of the few teams that has started to take this problem seriously.
Proof size matters because Aztec transactions travel through a peer-to-peer network. Every node that receives and forwards a proof pays its bandwidth cost. Aztec’s 20kb proofs are “ok”. If it ever becomes a bottleneck we can make them smaller with little effort because we use elliptic curves.
Post-quantum proof systems (the fast ones), on the other hand, have proofs of ~300KB, which is not ok.
Worse, there are zero post-quantum proving systems whose proofs are small enough to gossip across a p2p network whilst achieving acceptable transaction throughput. There are fundamental constraints which mean this is not likely to change!
Fast recursion offers another path. A node could receive two transaction proofs, produce one proof that establishes the validity of both, and forward the aggregate. Other nodes could repeat the process. Transaction data would still have to travel through the network, but the proof overhead could shrink across a batch.
Ethereum’s post-quantum research faces a related problem. Its proposed hash-based validator signatures lack BLS’s cheap native aggregation, so leanVM supplies a proof that many signatures pass verification. The draft EIP-8292 assigns that proving work to aggregators with suitable hardware. Their longer-term roadmap requires full virtual-machine proof aggregations which presents an extremely similar problem to ours.
For the Aztec approach to work on ordinary nodes, aggregation needs to be cheap. My working estimate is that two-to-one aggregation in roughly 20–100 milliseconds on a modern CPU could make it practical.
Nodes receive different transactions at different times, so they will duplicate work. An attacker may try to exploit that duplication. We need to establish how many aggregations one malicious transaction can induce and what happens to throughput under that load.
This problem is currently under-explored and poorly parametrised and substantial work needs to occur before a definitive solution can be decided on.
The main candidates are hash-based systems over binary fields and small prime fields, lattice-based systems, and proposals that combine both prime-field and binary-field representations.
The Flock paper gives a good example of the progress in binary-field proving. Its authors report proving 660,000 BLAKE3 hashes per second on an M4 mac on ten cores.
These are impressive numbers. Hashing is a major part of verifying hash-based proofs, so cheap hashing is good for recursion.
However, binary field systems are weak with regards to integer multiplication. Trustless access to Web2 data is one of the biggest untapped capabilities of blockchain. Private applications should be able to use passport signatures, credit credentials, TLS-based data proofs and signed web tokens without publishing the underlying information. That means dealing with ugly Web2 cryptography: RSA, large-field elliptic curves and integer arithmetic.
Binary-field arithmetic does not behave like integer arithmetic. In particular, integer addition has carries, and multiplication needs a representation and a circuit that handles them. A BLAKE3 benchmark tells us little about the cost of proving a 4096-bit RSA verification. We need to measure the operations our applications will use.
That is what makes hybrid approaches interesting to me: using one commitment both in a binary representation and a prime field representation depending on context. This is an active area of research with limited existing published work, which highlights how rapidly evolving this space is.
Lattice-based proving deserves serious attention too. The Jolt team’s September release replaced the elliptic-curve commitment scheme Dory with Akita, based on Module-SIS. The authors reported faster proving and verification and proofs below 100 KB. Their announcement also described zero knowledge as forthcoming, so those figures do not settle our comparison.
Lattice commitments retain algebraic structure that could be useful for the kinds of composition we rely on. That makes them attractive candidates to investigate, although translating CHONK’s techniques would require work. The field size is a design choice we need to evaluate within each construction; lattices do not impose one universal 128-bit-field requirement.
Much of the suspicion around lattice systems rests on vibes about whether the underlying structure is vulnerable to AI cryptoanalysis. There are good reasons to scrutinize a particular assumption, parameter set or reduction. “It uses lattices” does not supply that scrutiny.
Not all lattice crypto is equal. The fundamental short-integer-solution hardness assumption has had decades of cryptanalysis. Lattice crypto has a degraded reputation because some modern protocols layer on additional, weaker hardness assumptions that have shown to be broken by quantum adversaries. But there is no quantitative evidence to suspect the core harness assumptions of lattice crypto are vulnerable.
Either way, the rate of development gives me confidence that we will have good options. It also makes choosing a winner today premature. We need comparisons that include our applications, recursion, zero knowledge and security parameters together.
In Aztec transactions, digital signatures are private witnesses within private functions of smart contracts (zero-knowledge circuits), which helpfully means we don’t care so much about the size of an individual signature. This gives us more freedom than non-rollups e.g. Ethereum, where signature size directly contributes to their transaction size. This matters because post-quantum digital signatures are typically large.
Similar to other privacy protocols such as ZCash Sapling and Orchard, several Aztec keys are baked-into the preimage of an Aztec address. To enable ‘post-quantumification’ of the Aztec protocol, we’re exploring externalising those keys to a canonical keystore registry. An Aztec address would then cease to be an elliptic-curve public key, and instead would simply be a hash which serves as a lookup into this registry. The registry could store all manner of exotic post-quantum public keys for different use cases.
Aztec uses Poseidon2 throughout the protocol. Poseidon2 is technically post-quantum secure. And while Poseidon2 is secure for our chosen domain, recent breaks over small-field Poseidon signal that we should move to a different hash function in the long-term.
We’re confident that Poseidon2 will remain practically secure for many years, but the writing is on the wall and it would not be wise to use this hash function in Aztec’s future post-quantum system.
We will likely replace it with either Blake2s or Blake3. The Ethereum Foundation are also considering a canonical hash function for their post-quantum roadmap and we’ll more than likely follow their lead here. Binary-field proof systems can trivially handle BLAKE hashes so the need for Poseidon2 will have run its course.
Our proposed selection milestones are:
The encryption milestone needs circuit measurements before we settle the proving-system choice. The proving-system decision needs working recursion and zero knowledge, alongside representative application workloads. These dates give us a sequence for answering those questions.
After selection, we need to replace the cryptography across the system. That includes addresses and key derivation, application-specific key checks inside the kernels, account authentication, validator authentication and the proof chain that ends on Ethereum. We also need to handle existing private state and give users and applications a migration path.
Ethereum supplies another dependency. Aztec’s final proof verification and blob commitments sit within Ethereum’s cryptographic infrastructure. A full post-quantum migration has to account for Ethereum’s upgrades as well as our own.
The roughly three years between the final selection milestone and end of year 2030 leave room for integration to expose problems that isolated benchmarks cannot. Development and integration of multiple novel post-quantum cryptographic primitives into Aztec will be neither simple nor fast. In addition, the entire tech stack will have to be formally verified due to the incredible threat that AI poses with regards to finding and exploiting bugs.
I think Q-day within four years is unlikely, and the roadmap is designed under the assumption that post-quantum work is not the sole focus and obsession of the Aztec Foundation. If there is evidence of faster progress in quantum hardware that would bring forward Q-day, we will be able to accelerate our roadmap by diverting resources to this project.
Developers will not have to wait for that full migration to make better choices about private data. The January work will provide hardened libraries and documented patterns to build against. A developer building a private application will be able to protect the data they create today to stand the test of time.
The Aztec Network today activated Alpha V5, a major protocol upgrade passed by token-holder governance and executed onchain. Alpha V5 reduces private-transaction proving times by more than 2x compared to the previous version, lowers the cost of a fully private transaction by roughly 50%, resolves the critical issues found in V4, and sees the first wave of apps go live. Users can now send private transactions and earn yield on Aave simply by connecting their Ethereum wallets on Nyx, bridge from Ethereum to Aztec using Shield or TRAIN, privately collect NFTs on RavenHouse, or play Dark Forest Aztec, a hidden-information strategy game in a universe that lives entirely onchain.
"Alpha V5 continues Aztec's work at the frontier of client-side proving, with cryptographic breakthroughs that cut proving times by more than half this release," said Zac Williamson, Co-founder, Aztec Foundation. "We believe Aztec is now the fastest system in the world for proving a fully private transaction entirely on a user's own device, and every release moves the industry closer to private transactions at public transaction speeds."
As the only decentralized privacy L2, Aztec is the credibly neutral privacy layer for Ethereum. Aztec allows anyone to write smart contracts that include both private and public aspects – every private transaction is proven on the user's own device, so no operator, sequencer, or intermediary can see the data. The Alpha V5 proving improvements come from cryptographic advances that make this client-side proving faster than any prior release. The network remains in alpha, but with V5 it is ready for teams to begin building and deploying applications.
Making private transactions practical comes down to how quickly a proof can be generated on a user's own device, without offloading that work to a server that would learn what the user is doing. On Alpha V5, proving a private token transfer natively now takes approximately 2.5 seconds on a consumer laptop, down from 5.2 seconds on V4, and about 6.8 seconds in a browser, down from 12.5 seconds. Across every measured transaction flow, client-side proving times improved by approximately 2x compared with V4.
Alpha V5 lowers ECDSA signature-verification cost by approximately 2x, speeds up Poseidon2 hashing by approximately 3x, and reduces the protocol circuit gate count by approximately 50% (gate count is the number of individual operations a proving circuit must perform, and it is the main driver of how long a proof takes to generate). Each of these lowers the amount of work a device performs to prove a transaction, and the reduction in gate count in particular compounds across every proof the network generates.
Alpha V5 launches the first wave of apps on a network where privacy is built into the protocol rather than managed by an operator. On other networks that claim privacy, transactions still pass through an operator or node that reads them in plaintext, or depend on a viewing key that a third party holds, so users rely on someone else to protect their data and to decide when it gets disclosed. On Aztec, every private transaction is proven on the user's own device, so the app, the sequencer, and any operator never need to see the underlying data. Nyx is one of these apps, allowing users to privately send transactions and privately earn yield on Aave.
"On Ethereum, everything you do is public. That's why we built Nyx: a private account governed by your Ethereum wallet", said Nikhil, Co-founder of Nyx. "Now you can send, receive and earn in private. Nyx was the first app live on the Aztec Alpha, and we're excited to expand participation to more users with the added stability of Alpha V5."
Other apps on Alpha V5 include Azguard and Nethermind (wallets), Shield, TRAIN, and RavenHouse (bridges), and the Aztecscan block explorers. Also launching is Dark Forest Aztec, a game where users explore a universe, control planets, manage planetary energy, expand territory, and launch attacks through strategic play with private state and hidden actions.
Transaction fees on Aztec come from two main sources: the cost of proving a transaction and the cost of verifying the rollup proof on Ethereum. Alpha V5 reduces both. It lowers the network's proving-cost parameter by 50%, and it reduces the L1 gas required to verify a rollup proof by approximately 40%. Because rollup proofs are verified on Ethereum and that cost is shared across all transactions in a batch, the L1 reduction lowers fees for every user, while the lower proving-cost parameter reduces the per-transaction proving fee directly. Together, these bring the average cost of a fully private token transfer to under a $0.05 transaction cost.
Alpha V5 also hardens the network on several fronts. It resolves critical vulnerabilities found in Alpha V4 along with additional bugs discovered since launch. Aztec's bug bounty program on Cantina also drew more than 234 security researchers to participate. The network remains in alpha, and further bugs may surface as usage grows, but each release has closed the issues found in the last and strengthened the protocol against new ones. With the critical V4 issues resolved and these safeguards in place, Alpha V5 is stable enough for teams to begin building and deploying applications.
Alpha V5 is live now, view the Alpha V5 landing page for a full list of features, performance updates, and live apps to explore.
Aztec is the only decentralized, privacy-first Layer 2 on Ethereum. Developers write private and public logic in the same smart contract, and private functions are executed and proven on the user's own device, so no operator sees the underlying data. The protocol is upgraded through onchain governance, and the network settles to Ethereum. For more information, visit aztec.network.
Crypto is in a long night. It is no secret that the industry is facing challenging circumstances and there has been a clear consolidation of the industry. Right now we are seeing a focus on real traction, demonstrable value projects shipping practical solutions that will meaningfully reach users.
Some of that discipline is overdue. However, in times like these the properties that made crypto structurally different begin to look expendable. Decentralization slows you down. It makes upgrades harder. It makes institutional sales harder. It removes the control surfaces that the existing financial world knows how to buy.
We used to accept those costs as the price of building something durable. But, in a famine, they look like unaffordable affectations. Discarding them wholesale, however, is like selling the land out from under our feet.
Permissionless, uncensorable transaction networks with rich composability - this is the clay from which our industry was grown. The long term commercial health of our industry depends on preserving these properties in an age of privacy and institutional adoption.
These trade-offs become more challenging and pernicious when privacy is involved. Privacy is the narrative for crypto in 2026, and for good reason. It’s the missing piece that will deliver the traction and real use-cases that the industry so desperately needs.
The challenges of decentralization multiply under the constraints of privacy and what we are seeing in the industry is not a pivot, but a complete capitulation of all of the differentiable value that made crypto valuable.
I have spent nearly a decade building a network that marries programmable privacy with decentralization. A network where users keep their data, where applications are composable with one another, where transactions can settle without a privileged party learning everyone’s business or deciding which products are allowed to exist. That required new cryptography, new programming models, new state architecture, new wallets, and a fairly insane number of tradeoffs that are invisible until you try to build the thing yourself. There are easier products to ship.
A centralized privacy service can give institutions something legible quickly, replicating how the existing financial sector works: a responsible operator, a viewing key, a way to block transactions, a way to explain the whole thing to a risk committee. Some of these products will be useful. Some will be good businesses. But they are not the thing we came here to build.
Institutional and enterprise adoption is one of the core growth areas in this crypto-winter and the playbook is simple: use the language of crypto as a skin-suit to sell products and services that pattern match onto existing financial rails, with their need for complete visibility, censorship, centralized network operators and all of the liabilities this incurs.
This is a tempting bargain because it shortens the path to adoption. It gives buyers and regulators a shape they understand. A company. A contract. A switch. But the moment you accept that bargain, the system changes character. It may still be encrypted. It may still contain proofs. It may still call itself private. But, it now behaves like and is an operated service.
There is a party with privileged knowledge and privileged control. Builders must shape themselves around it. Institutions negotiate with it. Regulators may pressure it. Attackers target it. Users ultimately depend on it. By a backdoor I mean something specific: a network or protocol-level viewing key where the product developer does not control who can see their users’ data, especially when paired with network-level controls that can block transactions or ban smart contracts entirely. I do not mean application-level controls. I do not mean user-authorised disclosure. I do not mean a dapp deciding that users must prove something before using it. Regulated applications will need rules. The issue is that the disclosure boundary of your application belongs to somebody else, and the same layer that sees can also decide whether your users are allowed to transact. In short, users lack a platform that has credible neutrality.
Privacy on top of centralized rails is fatal. If one party can see everything and stop anything, that party may be treated as responsible for seeing and stopping.
This compounds into substantial platform risk. If an entity builds on top of such a system they must surrender visibility and control to the network operator to satisfy their liabilities without consideration for yours. Decentralization and ultimately credible neutrality is the difference between whether you own durable infrastructure or are renting a service whose rules can change on a whim. Worse, you cannot “just build things”. For novel transaction flows approval must be sought and granted. Tell me, would Ethereum have grown if every smart contract deployment required approval from the Ethereum Foundation?
Privacy needs the same freedom. A private credit market, for example, touches identity, collateral, repayment history, payment flows, liquidation logic, lender disclosures, auditor access and borrower privacy. If every component lives inside a different permissioned service, each with its own operator and viewing assumptions, that is a bureaucratic friction that negates blockchain’s core value proposition; composability.
A decentralized and credibly neutral privacy network prevents the settlement layer from becoming the single place where all surveillance and censorship obligations naturally accumulate. It allows product developers to scope their code to satisfy their own narrow requirements without consideration for the obligations of a centralized operator.
A lot of today’s privacy narrative treats architecture as if it were a detail. It is not. You cannot take a transparent ledger, staple confidentiality onto the edge, add a viewing key for comfort, and expect to get programmable private infrastructure.
If the state model is not private from the ground up you get wrappers, third party tools, data custodians, ad hoc disclosure paths and a pile of assumptions that every application drags into the next. Developers do not get a normal programming model where private contracts can call private contracts and users keep state on their own devices. They do not get composability.
The difference matters. In a real private execution environment, users generate transactions locally. They do not outsource their intent to a third party who learns what they are doing. Private contracts interact through a state model designed for privacy. The network settles proofs without becoming the party that knows everyone’s business. Privacy is part of the architecture.
This is why Aztec has taken so long. We built something that makes programmable private state and decentralised settlement live inside the same system. That means proving systems that run on consumer hardware, a transaction architecture built around local private execution, and a programming model where privacy is idiomatic and just works out of the box.
A centralized service can skip much of this. It can hold the key, run the prover, approve the flow and call the result privacy. It gets to market faster because it is not trying to arrive at the same place.
Adding decentralization does not make obligations disappear. Applications, issuers, frontends, custodians and regulated businesses will continue to exist in a web of obligations and responsibilities. Anyone pretending otherwise is unserious.
The question is where those obligations live. If they are pushed into the settlement layer, the settlement layer is no longer credibly neutral. It needs visibility into everyone and controls over everyone.
The better answer is selective disclosure. Users and applications should prove specific facts to specific parties for specific purposes. A regulated application may need to know that a user passed a check, that a transaction satisfies a policy, or that an auditor can inspect a particular flow. None of that requires the base network to hold a permanent key into everyone’s activity.
This will be harder to explain to the existing world. New infrastructure always fails to fit the categories built for the old infrastructure. Bitcoin did not arrive as a neatly regulated bank product. Ethereum did not wait for every lawyer to understand smart contracts. Stablecoins and DeFi forced institutions, regulators and users to develop new language around rails that kept existing.
If the standard for privacy infrastructure is to plug into the old world without changing anything, the answer will always be a service with a backdoor. And the result will be to catch crumbs falling from the tables of the old world.
The market we should be building is, well, a market. A private financial system that compounds: assets, liquidity, identity, credentials, credit and applications interacting through a shared settlement layer without forcing users to surrender their data to whoever sits in the middle.
Traditional finance is built out of vertically integrated information silos. Those silos are its moat. Banks, exchanges, custodians, payment processors and data brokers all benefit from controlling the information that flows through them. A global private settlement layer attacks that advantage directly. It lets liquidity and credentials move while outsourcing information custody to neutral cryptographic infrastructure.
A company wants a moat. A settlement layer wants surface area. A permissioned privacy provider can ration access, raise fees, exclude applications, shape disclosure rules and define acceptable use around its own risk tolerance. These are products pretending to be networks, and not durable financial infrastructure. What bothers me is this compounding category confusion. Networks adding protocol-level viewing keys and transaction controls are using the same language as decentralised programmable privacy, and commentators are treating them as variations of the same thing. They are not.
We have spent nine years walking the hard road. Now, just as we are close, the market has lost faith. Everyone is reaching for whatever lifeline looks immediate. Some of those lifelines will be real. Some will make money. But if crypto responds to its long night by rebuilding financial privacy as permissioned services, then we will have survived by surrendering the property that made the industry worth building.
Markets can grow when the platform is removed from the position where it can dictate the rules. It would be perverse to forget that lesson while building privacy, the domain where control over information matters most.
Crypto is in a famine. The land is struggling. We could sell our land for a pittance and survive the season. But the famine will pass, and when it does the land will blossom again. Without the land we are nothing.
We have struggled immensely to create a permissionless network that can marry privacy with decentralisation: an indestructible network whose users cannot be surveilled and whose transactions cannot be censored. This is the soil we have to grow our crops. To surrender a backdoor or a centralized operator for temporary relief is to sell our land for the price of a stablecoin. And we cannot sell the land.
Follow Zac on X to get more insights
Follow Aztec on X for updates & breaking news
Privacy has become a baseline requirement for L1s and L2s who care about bringing real-world users onchain. Users don't want their activity broadcast to competitors or the general public, but applications operating at scale also need some form of auditability, whether for regulators, compliance requirements, or tax reporting. Selective disclosure resolves that tension: privacy by default, with the ability to prove specific facts when required. What separates these networks is not whether they offer that switch, but who gets to hold it.
Aztec, Canton, Starknet, Tempo, and zkSync all offer some form of privacy with selective disclosure, but under the hood they make fundamentally different architectural decisions about who can see your data and who can turn your privacy off. Those decisions determine whether your privacy stays under your own control or sits behind a switch that someone else operates.
Three questions reveal where these networks actually diverge:
The answers determine whether your privacy off-switch is held by a policy, by an operator's good behavior, or by you alone through a cryptographic proof. As you'll see in this post, there are legitimate reasons to use each one with different tradeoffs. Aztec is the only network, however, where that switch stays in the user's hands, answering all three questions without putting a permissioned set of operators or a standing viewing key in control of your privacy. That gives developers the flexibility to build apps that comply with applicable laws while still keeping full privacy under the user's control.
This article will compare the privacy approaches of Aztec, Canton, Starknet, Tempo, and zkSync to give developers insight into the privacy tradeoffs of each network.
Here’s how each network handles the selective disclosure privacy off-switch, and who has control over your privacy:
Each of these networks offers privacy with selective disclosure, but each rests on a different network design with its own tradeoffs. We have ordered them by who holds your privacy off-switch, starting with designs where a third party controls access to your data and ending with designs where that control stays with you. At the top, the switch sits behind a policy promise and an honest operator, and further down it is replaced by proofs that the user generates and controls.
Canton keeps data private by controlling viewing permissions for the various actors on its network. A transaction splits into per-participant views, so each party receives only the sub-transactions that name it, and the parts it is not entitled to never reach it. The sequencer and mediator move those views without reading them, which is real privacy against those roles.
However, the data is still read in plaintext by the participant nodes that host the relevant parties, and in the common regulated-asset pattern where the issuer is a signatory on its own token, the issuer's node sees every transfer. The harder gap is verification, because no third party can reconstruct the global ledger, so correctness rests on the confirming nodes staying honest and their keys staying safe. In practice the off-switch sits with those nodes rather than with you, since you cannot see when your data is read and cannot stop it.
Tempo is designed for payments and uses validity proofs to verify that each zone is executing correctly, while still giving the zone operator full plaintext visibility into every transaction within that zone. Privacy comes from Tempo Zones, which are parallel execution environments connected to the Tempo mainnet.
By design, the zone operator has visibility into all transactions within the zone, while users see only their own and the public sees only a proof that the zone is valid. Token issuers set compliance controls, allowlists, blocklists, and freezes, enforced across zones. The mainnet checks each zone's validity, so execution is verified, while the operator still reads every transaction in plaintext and holds the off-switch over what is revealed. Your privacy is from the public, not from the operator.
zkSync Prividium adds the verifiability piece that Canton lacks. Every batch produces a validity proof settled to Ethereum, so a compromised operator cannot forge state or mint tokens from nothing without also forging a proof, which it cannot do. The tradeoff is that the operator processes every transaction in plaintext and decides who sees what, which means the off-switch stays with the operator and your privacy is from the outside world rather than from the operator itself.
This tradeoff has legitimate uses in high-trust institutional environments. If Bank of America, JPMorgan, and Wells Fargo are transacting on a shared network, a zone where BofA's infrastructure processes BofA-originated transactions satisfies internal control requirements while still delivering genuine ZK privacy from the other banks and the rest of the world. Where this model breaks down is in lower-trust environments where giving an operator full plaintext access and the switch that comes with it holds back product design possibilities.
Starknet's STRK20 breaks from relying on an operator for privacy. It shields ERC-20 balances and transfers in a privacy pool, and every private transaction carries a zero-knowledge proof generated client-side, so no operator sees your plaintext in order to build it.
Disclosure is where STRK20 diverges from Aztec. To join the Starknet Privacy Pool, you register an encrypted viewing key onchain, and it sits there for the life of your participation. On a regulatory request, a designated auditing entity can decrypt that key and trace your complete transaction history, forwards and backwards. StarkWare calls this ‘not a backdoor’ but a carefully scoped access mechanism, and the safeguard is a policy promise that the auditor decrypts only when required. The privacy is cryptographic, but the off-switch is a standing key that someone else holds and can flip whether or not you are watching.
On Aztec your private state lives as encrypted private data that only you can decrypt. The contract developer can choose what state is public and what is private, and whether your encrypted private data is emitted onchain as a private log or shared off-chain instead.
Your transactions get proven client-side on your own device, so no sequencer or operator sees your unencrypted private data. Those proofs settle to Ethereum, which gives the same integrity anchor marketed by Prividium, with every transaction verified and no forged state, but without a single operator who reads your data. The base protocol decentralizes sequencing, proving, and governance, so there is no operator to choose and trust in the first place.
Disclosure is your choice too: you decide who learns your private data, and whether they learn it in encrypted or decrypted form. To grant discovery without readability, you share an app-specific tagging secret that lets an auditor find your data in encrypted form without being able to decrypt and read it. This is enough to prove things calculated from that data, such as a tax basis or a profit and loss figure. Granting permission to actually read the data works differently. There's no per-contract read key you can hand out, because decryption uses your master viewing key, which would unlock all your data across every contract. So instead of sharing a key, you share the data itself, plus a proof that your plaintext is what encrypts to the on-chain ciphertext.
Aztec has true selective disclosure in that you can selectively share it, and nothing else you don’t need to. This is app specific, meaning that private data discoverability access on one app does not grant access on another. Most importantly, the off-switch stays in your hands, and you never need to trust the network to handle access to any of your private data and activity.
This is not just conceptual: here is a working proof-of-concept of this model on Aztec. PrivPNL takes you from private DEX trades through a tagging-key disclosure to a browser-generated ZK proof of your PnL. The auditor verifies a proof while the prover only has to reveal the amount they owe, and your portfolio stays private.
Canton keeps the switch with the participant nodes that read your data in plaintext, so disclosure rests on those nodes staying honest rather than on anything you control. Tempo similarly gives the off-switch to a zone-based node operator, but allows you to verify the correctness of transactions using validity proofs. Prividium hardens that promise with a proof settled to Ethereum, a real improvement, but the operator still reads every transaction and still decides who sees what. This can work well for large institutions, but small to medium sized enterprises are left with the same privacy as their current banks unless they run their own Prividium nodes. STRK20 moves the switch into a standing viewing key and asks you to trust that a designated auditor reaches for it only when needed. In each of these models the real question is not whether your privacy can be switched off, but who gets to do the switching, and whether you would even know it happened.
Aztec takes the operator and the standing key out of the question entirely. You keep the data, you generate the proof, and you disclose the result, one fact at a time and only when you choose to. The off-switch never leaves your hands, and no operator, auditor, or node can reach it on your behalf. This is one of the benefits of a network that offers fully programmable, privacy-preserving smart contracts that put you in control.
Selective disclosure is how privacy survives contact with a regulator, and the model you pick decides who can open your history when you are not looking. On Aztec, that answer is no one but you.
Dive into the technical details: Try a live demo of selective disclosure on Aztec and read the technical article on how it was built.
Integrate with Aztec: Reach out if you are interested in integrating privacy into your project.
The Aztec Network today activated Alpha V5, a major protocol upgrade passed by token-holder governance and executed onchain. Alpha V5 reduces private-transaction proving times by more than 2x compared to the previous version, lowers the cost of a fully private transaction by roughly 50%, resolves the critical issues found in V4, and sees the first wave of apps go live. Users can now send private transactions and earn yield on Aave simply by connecting their Ethereum wallets on Nyx, bridge from Ethereum to Aztec using Shield or TRAIN, privately collect NFTs on RavenHouse, or play Dark Forest Aztec, a hidden-information strategy game in a universe that lives entirely onchain.
"Alpha V5 continues Aztec's work at the frontier of client-side proving, with cryptographic breakthroughs that cut proving times by more than half this release," said Zac Williamson, Co-founder, Aztec Foundation. "We believe Aztec is now the fastest system in the world for proving a fully private transaction entirely on a user's own device, and every release moves the industry closer to private transactions at public transaction speeds."
As the only decentralized privacy L2, Aztec is the credibly neutral privacy layer for Ethereum. Aztec allows anyone to write smart contracts that include both private and public aspects – every private transaction is proven on the user's own device, so no operator, sequencer, or intermediary can see the data. The Alpha V5 proving improvements come from cryptographic advances that make this client-side proving faster than any prior release. The network remains in alpha, but with V5 it is ready for teams to begin building and deploying applications.
Making private transactions practical comes down to how quickly a proof can be generated on a user's own device, without offloading that work to a server that would learn what the user is doing. On Alpha V5, proving a private token transfer natively now takes approximately 2.5 seconds on a consumer laptop, down from 5.2 seconds on V4, and about 6.8 seconds in a browser, down from 12.5 seconds. Across every measured transaction flow, client-side proving times improved by approximately 2x compared with V4.
Alpha V5 lowers ECDSA signature-verification cost by approximately 2x, speeds up Poseidon2 hashing by approximately 3x, and reduces the protocol circuit gate count by approximately 50% (gate count is the number of individual operations a proving circuit must perform, and it is the main driver of how long a proof takes to generate). Each of these lowers the amount of work a device performs to prove a transaction, and the reduction in gate count in particular compounds across every proof the network generates.
Alpha V5 launches the first wave of apps on a network where privacy is built into the protocol rather than managed by an operator. On other networks that claim privacy, transactions still pass through an operator or node that reads them in plaintext, or depend on a viewing key that a third party holds, so users rely on someone else to protect their data and to decide when it gets disclosed. On Aztec, every private transaction is proven on the user's own device, so the app, the sequencer, and any operator never need to see the underlying data. Nyx is one of these apps, allowing users to privately send transactions and privately earn yield on Aave.
"On Ethereum, everything you do is public. That's why we built Nyx: a private account governed by your Ethereum wallet", said Nikhil, Co-founder of Nyx. "Now you can send, receive and earn in private. Nyx was the first app live on the Aztec Alpha, and we're excited to expand participation to more users with the added stability of Alpha V5."
Other apps on Alpha V5 include Azguard and Nethermind (wallets), Shield, TRAIN, and RavenHouse (bridges), and the Aztecscan block explorers. Also launching is Dark Forest Aztec, a game where users explore a universe, control planets, manage planetary energy, expand territory, and launch attacks through strategic play with private state and hidden actions.
Transaction fees on Aztec come from two main sources: the cost of proving a transaction and the cost of verifying the rollup proof on Ethereum. Alpha V5 reduces both. It lowers the network's proving-cost parameter by 50%, and it reduces the L1 gas required to verify a rollup proof by approximately 40%. Because rollup proofs are verified on Ethereum and that cost is shared across all transactions in a batch, the L1 reduction lowers fees for every user, while the lower proving-cost parameter reduces the per-transaction proving fee directly. Together, these bring the average cost of a fully private token transfer to under a $0.05 transaction cost.
Alpha V5 also hardens the network on several fronts. It resolves critical vulnerabilities found in Alpha V4 along with additional bugs discovered since launch. Aztec's bug bounty program on Cantina also drew more than 234 security researchers to participate. The network remains in alpha, and further bugs may surface as usage grows, but each release has closed the issues found in the last and strengthened the protocol against new ones. With the critical V4 issues resolved and these safeguards in place, Alpha V5 is stable enough for teams to begin building and deploying applications.
Alpha V5 is live now, view the Alpha V5 landing page for a full list of features, performance updates, and live apps to explore.
Aztec is the only decentralized, privacy-first Layer 2 on Ethereum. Developers write private and public logic in the same smart contract, and private functions are executed and proven on the user's own device, so no operator sees the underlying data. The protocol is upgraded through onchain governance, and the network settles to Ethereum. For more information, visit aztec.network.
On Ethereum today, each transaction reveals everything publicly. The token you moved, the size, the timing, the wallet it came from, every action you take. Given the limitations of this type of transparent network, the industry is now focusing on bringing privacy onchain as a top priority. The response to this has mostly been to enable private transactions that shield transfers in various ways. But when we look at how privacy works on Web2, it’s clear that users and developers need granular privacy controls: the ability to decide what is public or private and who is able to see different types of data.
Aztec was built so that one transaction can carry two halves. A private half that runs on your own device and never leaves it, and a public half that the network runs in the open. Apps can choose which aspects are private or public, and users can choose what they want to reveal and when.
This article will follow an example transaction on Aztec: a vote in an onchain election built on Aztec, where who you are and which candidate you chose stay private, while the running tally for each candidate stays public for anyone to verify.
Picture the vote you cast in our example as two aspects that seamlessly weave together. In the first step, you act in private: an app records your vote on your device and hands the network a proof that the vote is valid without revealing it. In the second, the network acts in public: it checks that proof, then adds one to the chosen candidate's public tally. It is one transaction: one part stays with you, one part goes to the network. Both parts end up recorded onchain, in two separate state trees, one private and one public. The walkthrough below follows how these two aspects work together and what this means for how your transaction lands onchain.
You open the voting app and connect an Aztec wallet. That first step looks like any onchain app. The difference is inside the wallet. An Aztec wallet carries a private execution environment, the PXE, pronounced "pixie", which runs on your phone or in your browser. The PXE is where the private half of your transaction executes, and where the proof of that work gets made, on your hardware, under your exclusive control.
Every account on Aztec is a smart contract rather than a bare key. That design, account abstraction, allows a wallet to authorize a transaction however its owner chooses without writing an identity onto the network for everyone to read. The wallet is the front door, and on Aztec you can decide if the door is open or closed, who you share your information with.
The voting app is a smart contract with two kinds of functions. The private functions run first, and they run inside your PXE. Your identity and the candidate you picked are the private inputs, and they stay on your device.
The only thing to leave your device is a proof confirming the legitimacy of your vote. Aztec's client-side proving system, Chonk, takes the private execution and produces a zero-knowledge proof: a compact cryptographic receipt that your vote followed the rules, that you are eligible, and have not voted before, while revealing nothing about who you are or who you voted for. Think of it as a sealed ballot the network can confirm is valid without opening it. The network learns only that a legitimate vote happened. It does not learn how you voted, or even which account voted.
This is the part that used to be too slow to be practical. Generating a proof on a phone was the bottleneck every privacy app hit. Aztec’s Chonk is purpose-built for fast proving on low-memory devices, both natively and in the browser, so the private half runs on the device in your hand instead of on someone else's server.
Some elements of a vote should be public. The tally is shared infrastructure, the number everyone relies on to trust the result. Thanks to programmable privacy on Aztec, the app marks that part public. Public functions live on the network and run in the open, the way functions do on Ethereum.
On Aztec, private and public logic live in the same contract, and the developer decides which is which, function by function and variable by variable. Programmable privacy is a dimmer, not a switch. The voting app turns it up on the individual ballot and turns it down on the running tally. That boundary is a design decision written into the contract, and it is the thing no transparent chain and no fixed-privacy chain can offer.
Your vote leaves your device as a bundle: the zero-knowledge proof of the private half, plus the call to the public function that updates the count. It goes to Aztec's sequencers, a decentralized set of thousands of independent operators, with more than 3,500 of them running the network today.
The sequencers do two jobs at once. They verify the proof of your private vote, confirming it is valid and eligible without seeing the choice behind it, and they run the public function that adds one to the chosen candidate and updates the public tally. Your ballot stays sealed. The count goes up by one for everyone to see. The same proof guarantees you cannot vote twice, even though no one learns which ballot is yours.
Aztec has two main state trees, and both live onchain. One holds private state, the other holds public state, so the full record of what happened sits on the network rather than on any one person's laptop. The two trees store each record in two different ways depending on if it needs to be private or public.
The private tree uses a UTXO model, the same note-based design used by Zcash. In this model, state is written as commitments: each entry is a sealed record that a valid vote was cast, with the voter and the choice kept private. Just like with Zcash or Bitcoin, you do not edit a private entry in place. You write a new one, and the design stops the same vote from being cast twice (old state is nullified). The vote stays private, and the record of a legitimate vote happening is onchain for the network to check.
The public tree uses an account-based model, the same shape Ethereum uses: values that update in place, readable by anyone. This is where each candidate's tally lives.
One transaction wrote information to both trees. The private tree recorded that you voted, sealed. The public tree recorded the new totals, in the open. Everything is onchain. The difference between the two trees is how much each one reveals.
Every private app on Aztec writes into that same private tree. A vote, a payment, and a payroll run all land in one shared record of activity, so each user's privacy grows stronger as the network grows, instead of splitting into a separate pool for every app.
Aztec is an L2 on Ethereum, so everything settles to Ethereum L1. A sequencer on Aztec gathers transactions into a proposed block. Other sequencers validate it before it goes to Ethereum's pending chain. At that point the block sits on Ethereum, ordered and recorded, waiting for its proof. The network has agreed on what happened and the proposed block is just waiting a final proof.
Proving a block is its own job, and on Aztec, it belongs to no one in particular. A decentralized, permissionless set of provers competes to take a full epoch, a 32-block stretch of the chain, and compresses it into a single zero-knowledge proof of the entire epoch. Anyone with the hardware can run a prover and bid for the work. There is no privileged operator, no committee you have to trust, no outside network holding a key.
That openness is the whole point of a privacy layer. A system that protects your data but routes it through one trusted server has only moved the exposure rather than removed it. Aztec keeps proving permissionless and your private inputs on your device, thereby avoiding any exposure.
The economics land in the voter's favor too. As an L2 network, Aztec spreads the cost of that one L1 proof across thousands of transactions in the rollup, so a vote costs pennies, not the millions of gas a private proof would cost verified alone on Ethereum.
A prover then posts the epoch proof to Ethereum's proven chain, and the Aztec state is final. Ethereum verifies one proof and inherits the correctness of everything inside it. Aztec extends Ethereum and settles to Ethereum, so your hybrid transaction carries Ethereum's security without carrying Ethereum's enforced transparency.
Anyone can now verify that the result is valid and that every counted vote was legitimate. No one can see how any individual voted. The tally is on the shared ledger where it belongs, and your ballot stayed yours the whole way through.
For the voter, their ballot was never a broadcast. The candidate you chose stayed yours, with no record tying your wallet to a name for anyone to read later, and you can still check that your vote was counted and the result is honest. You took part without your choice becoming data for systems built to act on it.
For a founder, the election app in this walkthrough is easy to implement without needing to build extensive custom code. Secret ballots with a public, verifiable count, in one contract, is a product category that opens up only because the boundary is programmable. You can build governance, elections, and polls where people vote without fear and the result still proves itself. And of course you can build anything that requires both public and private state to work seamlessly together.
For an infrastructure provider, the same machinery serves clients who need a result they can stand behind without exposing the people who produced it. Selective disclosure lets a client prove exactly what a counterparty needs to see, the count and the integrity of the process, and protect everything else, on their own terms. That is a guarantee a transparent chain cannot make.
A real vote needs two things at once: a secret ballot and a count anyone can check. A transparent chain makes you give up the first to get the second. On Aztec, you get both. The tally settled on Ethereum for anyone to verify, and how you voted stayed yours. The infrastructure is in place, what will you create with it?
Crypto is in a long night. It is no secret that the industry is facing challenging circumstances and there has been a clear consolidation of the industry. Right now we are seeing a focus on real traction, demonstrable value projects shipping practical solutions that will meaningfully reach users.
Some of that discipline is overdue. However, in times like these the properties that made crypto structurally different begin to look expendable. Decentralization slows you down. It makes upgrades harder. It makes institutional sales harder. It removes the control surfaces that the existing financial world knows how to buy.
We used to accept those costs as the price of building something durable. But, in a famine, they look like unaffordable affectations. Discarding them wholesale, however, is like selling the land out from under our feet.
Permissionless, uncensorable transaction networks with rich composability - this is the clay from which our industry was grown. The long term commercial health of our industry depends on preserving these properties in an age of privacy and institutional adoption.
These trade-offs become more challenging and pernicious when privacy is involved. Privacy is the narrative for crypto in 2026, and for good reason. It’s the missing piece that will deliver the traction and real use-cases that the industry so desperately needs.
The challenges of decentralization multiply under the constraints of privacy and what we are seeing in the industry is not a pivot, but a complete capitulation of all of the differentiable value that made crypto valuable.
I have spent nearly a decade building a network that marries programmable privacy with decentralization. A network where users keep their data, where applications are composable with one another, where transactions can settle without a privileged party learning everyone’s business or deciding which products are allowed to exist. That required new cryptography, new programming models, new state architecture, new wallets, and a fairly insane number of tradeoffs that are invisible until you try to build the thing yourself. There are easier products to ship.
A centralized privacy service can give institutions something legible quickly, replicating how the existing financial sector works: a responsible operator, a viewing key, a way to block transactions, a way to explain the whole thing to a risk committee. Some of these products will be useful. Some will be good businesses. But they are not the thing we came here to build.
Institutional and enterprise adoption is one of the core growth areas in this crypto-winter and the playbook is simple: use the language of crypto as a skin-suit to sell products and services that pattern match onto existing financial rails, with their need for complete visibility, censorship, centralized network operators and all of the liabilities this incurs.
This is a tempting bargain because it shortens the path to adoption. It gives buyers and regulators a shape they understand. A company. A contract. A switch. But the moment you accept that bargain, the system changes character. It may still be encrypted. It may still contain proofs. It may still call itself private. But, it now behaves like and is an operated service.
There is a party with privileged knowledge and privileged control. Builders must shape themselves around it. Institutions negotiate with it. Regulators may pressure it. Attackers target it. Users ultimately depend on it. By a backdoor I mean something specific: a network or protocol-level viewing key where the product developer does not control who can see their users’ data, especially when paired with network-level controls that can block transactions or ban smart contracts entirely. I do not mean application-level controls. I do not mean user-authorised disclosure. I do not mean a dapp deciding that users must prove something before using it. Regulated applications will need rules. The issue is that the disclosure boundary of your application belongs to somebody else, and the same layer that sees can also decide whether your users are allowed to transact. In short, users lack a platform that has credible neutrality.
Privacy on top of centralized rails is fatal. If one party can see everything and stop anything, that party may be treated as responsible for seeing and stopping.
This compounds into substantial platform risk. If an entity builds on top of such a system they must surrender visibility and control to the network operator to satisfy their liabilities without consideration for yours. Decentralization and ultimately credible neutrality is the difference between whether you own durable infrastructure or are renting a service whose rules can change on a whim. Worse, you cannot “just build things”. For novel transaction flows approval must be sought and granted. Tell me, would Ethereum have grown if every smart contract deployment required approval from the Ethereum Foundation?
Privacy needs the same freedom. A private credit market, for example, touches identity, collateral, repayment history, payment flows, liquidation logic, lender disclosures, auditor access and borrower privacy. If every component lives inside a different permissioned service, each with its own operator and viewing assumptions, that is a bureaucratic friction that negates blockchain’s core value proposition; composability.
A decentralized and credibly neutral privacy network prevents the settlement layer from becoming the single place where all surveillance and censorship obligations naturally accumulate. It allows product developers to scope their code to satisfy their own narrow requirements without consideration for the obligations of a centralized operator.
A lot of today’s privacy narrative treats architecture as if it were a detail. It is not. You cannot take a transparent ledger, staple confidentiality onto the edge, add a viewing key for comfort, and expect to get programmable private infrastructure.
If the state model is not private from the ground up you get wrappers, third party tools, data custodians, ad hoc disclosure paths and a pile of assumptions that every application drags into the next. Developers do not get a normal programming model where private contracts can call private contracts and users keep state on their own devices. They do not get composability.
The difference matters. In a real private execution environment, users generate transactions locally. They do not outsource their intent to a third party who learns what they are doing. Private contracts interact through a state model designed for privacy. The network settles proofs without becoming the party that knows everyone’s business. Privacy is part of the architecture.
This is why Aztec has taken so long. We built something that makes programmable private state and decentralised settlement live inside the same system. That means proving systems that run on consumer hardware, a transaction architecture built around local private execution, and a programming model where privacy is idiomatic and just works out of the box.
A centralized service can skip much of this. It can hold the key, run the prover, approve the flow and call the result privacy. It gets to market faster because it is not trying to arrive at the same place.
Adding decentralization does not make obligations disappear. Applications, issuers, frontends, custodians and regulated businesses will continue to exist in a web of obligations and responsibilities. Anyone pretending otherwise is unserious.
The question is where those obligations live. If they are pushed into the settlement layer, the settlement layer is no longer credibly neutral. It needs visibility into everyone and controls over everyone.
The better answer is selective disclosure. Users and applications should prove specific facts to specific parties for specific purposes. A regulated application may need to know that a user passed a check, that a transaction satisfies a policy, or that an auditor can inspect a particular flow. None of that requires the base network to hold a permanent key into everyone’s activity.
This will be harder to explain to the existing world. New infrastructure always fails to fit the categories built for the old infrastructure. Bitcoin did not arrive as a neatly regulated bank product. Ethereum did not wait for every lawyer to understand smart contracts. Stablecoins and DeFi forced institutions, regulators and users to develop new language around rails that kept existing.
If the standard for privacy infrastructure is to plug into the old world without changing anything, the answer will always be a service with a backdoor. And the result will be to catch crumbs falling from the tables of the old world.
The market we should be building is, well, a market. A private financial system that compounds: assets, liquidity, identity, credentials, credit and applications interacting through a shared settlement layer without forcing users to surrender their data to whoever sits in the middle.
Traditional finance is built out of vertically integrated information silos. Those silos are its moat. Banks, exchanges, custodians, payment processors and data brokers all benefit from controlling the information that flows through them. A global private settlement layer attacks that advantage directly. It lets liquidity and credentials move while outsourcing information custody to neutral cryptographic infrastructure.
A company wants a moat. A settlement layer wants surface area. A permissioned privacy provider can ration access, raise fees, exclude applications, shape disclosure rules and define acceptable use around its own risk tolerance. These are products pretending to be networks, and not durable financial infrastructure. What bothers me is this compounding category confusion. Networks adding protocol-level viewing keys and transaction controls are using the same language as decentralised programmable privacy, and commentators are treating them as variations of the same thing. They are not.
We have spent nine years walking the hard road. Now, just as we are close, the market has lost faith. Everyone is reaching for whatever lifeline looks immediate. Some of those lifelines will be real. Some will make money. But if crypto responds to its long night by rebuilding financial privacy as permissioned services, then we will have survived by surrendering the property that made the industry worth building.
Markets can grow when the platform is removed from the position where it can dictate the rules. It would be perverse to forget that lesson while building privacy, the domain where control over information matters most.
Crypto is in a famine. The land is struggling. We could sell our land for a pittance and survive the season. But the famine will pass, and when it does the land will blossom again. Without the land we are nothing.
We have struggled immensely to create a permissionless network that can marry privacy with decentralisation: an indestructible network whose users cannot be surveilled and whose transactions cannot be censored. This is the soil we have to grow our crops. To surrender a backdoor or a centralized operator for temporary relief is to sell our land for the price of a stablecoin. And we cannot sell the land.
Follow Zac on X to get more insights
Follow Aztec on X for updates & breaking news
Privacy has become a baseline requirement for L1s and L2s who care about bringing real-world users onchain. Users don't want their activity broadcast to competitors or the general public, but applications operating at scale also need some form of auditability, whether for regulators, compliance requirements, or tax reporting. Selective disclosure resolves that tension: privacy by default, with the ability to prove specific facts when required. What separates these networks is not whether they offer that switch, but who gets to hold it.
Aztec, Canton, Starknet, Tempo, and zkSync all offer some form of privacy with selective disclosure, but under the hood they make fundamentally different architectural decisions about who can see your data and who can turn your privacy off. Those decisions determine whether your privacy stays under your own control or sits behind a switch that someone else operates.
Three questions reveal where these networks actually diverge:
The answers determine whether your privacy off-switch is held by a policy, by an operator's good behavior, or by you alone through a cryptographic proof. As you'll see in this post, there are legitimate reasons to use each one with different tradeoffs. Aztec is the only network, however, where that switch stays in the user's hands, answering all three questions without putting a permissioned set of operators or a standing viewing key in control of your privacy. That gives developers the flexibility to build apps that comply with applicable laws while still keeping full privacy under the user's control.
This article will compare the privacy approaches of Aztec, Canton, Starknet, Tempo, and zkSync to give developers insight into the privacy tradeoffs of each network.
Here’s how each network handles the selective disclosure privacy off-switch, and who has control over your privacy:
Each of these networks offers privacy with selective disclosure, but each rests on a different network design with its own tradeoffs. We have ordered them by who holds your privacy off-switch, starting with designs where a third party controls access to your data and ending with designs where that control stays with you. At the top, the switch sits behind a policy promise and an honest operator, and further down it is replaced by proofs that the user generates and controls.
Canton keeps data private by controlling viewing permissions for the various actors on its network. A transaction splits into per-participant views, so each party receives only the sub-transactions that name it, and the parts it is not entitled to never reach it. The sequencer and mediator move those views without reading them, which is real privacy against those roles.
However, the data is still read in plaintext by the participant nodes that host the relevant parties, and in the common regulated-asset pattern where the issuer is a signatory on its own token, the issuer's node sees every transfer. The harder gap is verification, because no third party can reconstruct the global ledger, so correctness rests on the confirming nodes staying honest and their keys staying safe. In practice the off-switch sits with those nodes rather than with you, since you cannot see when your data is read and cannot stop it.
Tempo is designed for payments and uses validity proofs to verify that each zone is executing correctly, while still giving the zone operator full plaintext visibility into every transaction within that zone. Privacy comes from Tempo Zones, which are parallel execution environments connected to the Tempo mainnet.
By design, the zone operator has visibility into all transactions within the zone, while users see only their own and the public sees only a proof that the zone is valid. Token issuers set compliance controls, allowlists, blocklists, and freezes, enforced across zones. The mainnet checks each zone's validity, so execution is verified, while the operator still reads every transaction in plaintext and holds the off-switch over what is revealed. Your privacy is from the public, not from the operator.
zkSync Prividium adds the verifiability piece that Canton lacks. Every batch produces a validity proof settled to Ethereum, so a compromised operator cannot forge state or mint tokens from nothing without also forging a proof, which it cannot do. The tradeoff is that the operator processes every transaction in plaintext and decides who sees what, which means the off-switch stays with the operator and your privacy is from the outside world rather than from the operator itself.
This tradeoff has legitimate uses in high-trust institutional environments. If Bank of America, JPMorgan, and Wells Fargo are transacting on a shared network, a zone where BofA's infrastructure processes BofA-originated transactions satisfies internal control requirements while still delivering genuine ZK privacy from the other banks and the rest of the world. Where this model breaks down is in lower-trust environments where giving an operator full plaintext access and the switch that comes with it holds back product design possibilities.
Starknet's STRK20 breaks from relying on an operator for privacy. It shields ERC-20 balances and transfers in a privacy pool, and every private transaction carries a zero-knowledge proof generated client-side, so no operator sees your plaintext in order to build it.
Disclosure is where STRK20 diverges from Aztec. To join the Starknet Privacy Pool, you register an encrypted viewing key onchain, and it sits there for the life of your participation. On a regulatory request, a designated auditing entity can decrypt that key and trace your complete transaction history, forwards and backwards. StarkWare calls this ‘not a backdoor’ but a carefully scoped access mechanism, and the safeguard is a policy promise that the auditor decrypts only when required. The privacy is cryptographic, but the off-switch is a standing key that someone else holds and can flip whether or not you are watching.
On Aztec your private state lives as encrypted private data that only you can decrypt. The contract developer can choose what state is public and what is private, and whether your encrypted private data is emitted onchain as a private log or shared off-chain instead.
Your transactions get proven client-side on your own device, so no sequencer or operator sees your unencrypted private data. Those proofs settle to Ethereum, which gives the same integrity anchor marketed by Prividium, with every transaction verified and no forged state, but without a single operator who reads your data. The base protocol decentralizes sequencing, proving, and governance, so there is no operator to choose and trust in the first place.
Disclosure is your choice too: you decide who learns your private data, and whether they learn it in encrypted or decrypted form. To grant discovery without readability, you share an app-specific tagging secret that lets an auditor find your data in encrypted form without being able to decrypt and read it. This is enough to prove things calculated from that data, such as a tax basis or a profit and loss figure. Granting permission to actually read the data works differently. There's no per-contract read key you can hand out, because decryption uses your master viewing key, which would unlock all your data across every contract. So instead of sharing a key, you share the data itself, plus a proof that your plaintext is what encrypts to the on-chain ciphertext.
Aztec has true selective disclosure in that you can selectively share it, and nothing else you don’t need to. This is app specific, meaning that private data discoverability access on one app does not grant access on another. Most importantly, the off-switch stays in your hands, and you never need to trust the network to handle access to any of your private data and activity.
This is not just conceptual: here is a working proof-of-concept of this model on Aztec. PrivPNL takes you from private DEX trades through a tagging-key disclosure to a browser-generated ZK proof of your PnL. The auditor verifies a proof while the prover only has to reveal the amount they owe, and your portfolio stays private.
Canton keeps the switch with the participant nodes that read your data in plaintext, so disclosure rests on those nodes staying honest rather than on anything you control. Tempo similarly gives the off-switch to a zone-based node operator, but allows you to verify the correctness of transactions using validity proofs. Prividium hardens that promise with a proof settled to Ethereum, a real improvement, but the operator still reads every transaction and still decides who sees what. This can work well for large institutions, but small to medium sized enterprises are left with the same privacy as their current banks unless they run their own Prividium nodes. STRK20 moves the switch into a standing viewing key and asks you to trust that a designated auditor reaches for it only when needed. In each of these models the real question is not whether your privacy can be switched off, but who gets to do the switching, and whether you would even know it happened.
Aztec takes the operator and the standing key out of the question entirely. You keep the data, you generate the proof, and you disclose the result, one fact at a time and only when you choose to. The off-switch never leaves your hands, and no operator, auditor, or node can reach it on your behalf. This is one of the benefits of a network that offers fully programmable, privacy-preserving smart contracts that put you in control.
Selective disclosure is how privacy survives contact with a regulator, and the model you pick decides who can open your history when you are not looking. On Aztec, that answer is no one but you.
Dive into the technical details: Try a live demo of selective disclosure on Aztec and read the technical article on how it was built.
Integrate with Aztec: Reach out if you are interested in integrating privacy into your project.
