Research
8 Nov
## min read

Fully Confidential Ethereum Transactions: Aztec Network’s Privacy Architecture

Explore Aztec's groundbreaking architecture for Ethereum transactions, ensuring absolute privacy and setting a new standard in blockchain confidentiality.

Share
Written by
Jon Wu
Edited by

Aztec is a privacy-first zero-knowledge rollup on Ethereum: that means it’s the only Layer 2 built from the ground up to be fully privacy preserving.

To understand the paradigm-changing nature of private transactions and why it’s important to build privacy directly into a network’s architecture, we have to first discuss why Ethereum is not private.

Ethereum: A Public Blockchain

You might’ve heard of the term public ledger, which consists of two parts: accounts and balances.

The most primitive transaction on Ethereum is sending Ether from one account (address) to another. The way the network keeps track of this is by incrementing one account’s balance and decrementing the other’s — in other words, the ETH doesn’t really “move” in any sense.

Let’s look at an example transaction in detail: say my man snoopdogg.eth wants to send a transaction to cozomomedici.eth.

Just two businessmen.

Here’s how it shakes out: Snoop starts with 100 ETH and his account is debited 20 ETH. Cozomo starts with 0 ETH and his account is credited 20 ETH. Snoop’s ending account balance is 80 ETH. Cozomo’s is 20 ETH. Transfer complete.

An accounting ledger representation of a simple ETH transfer.

We can see a representation of credits and debits for each account right on etherscan.io, with the “ins” and “outs” tracked in public for everyone to see. Here’s the recent transaction history for an ENS named twinkienft.eth (a name I quite like):

Here it is in all its glory: twinkienft.eth’s public transactions!

You might be wondering: “Who’s twinkienft.eth?” I don’t have a clue, but I can see all their transactions! If you go to etherscan.io you can witness all transactions being written to the blockchain.

0x9dae… right on the front-page of etherscan.io!

You can see the obvious problem here. Not only can we see all account transactions, we can see all the amounts, assets, and counterparties.

That’s in fact the power of public blockchains! Due to their public nature they are eminently auditable and verifiable.

But that means if someone’s privacy is compromised, whether intentionally or by accident — we know their entire transaction history.

Cracking the public transaction graph is big business: companies like Chainalysis and Nansen run sophisticated forensic analysis to associate various wallets, monitor activity, and make probabilistic assumptions about who owns what.

Imagine if every time you swiped a credit card to buy a croissant you showed every person in the world your bank statement. That’d be, like, pretty goofy, right?

That’s the state of Ethereum today.

The Obvious Answer: Encrypted Accounts

“Uhm, okay,” I hear you saying. “This is so easy to solve, just encrypt the accounts, balances, and owners.” Duh, idiot! How could I be so stupid.

How I feel pretty much every day.

Except let’s actually talk through how encrypted accounts would work:

Recall the ledger from before. With encrypted accounts and transactions, it would instead look like this:

Useful.

ow would the network check the accounting, ensuring no double spend or collusive funny business? It turns out solving this is pretty f-ing hard!

Back to our businessmen Snoop & Cozomo to help us figure it out. If they need to do a transaction, they’ll have to interact, since the network can’t help check that they did a valid transaction.

If it did, someone somewhere would have knowledge of what went down. Instead, Snoop initiates the interaction:

  1. Snoop requests Cozomo’s encrypted account state
  2. Cozomo sends the encrypted state to Snoop
  3. Snoop decrypts Cozomo’s state, confirming the pre-transaction balance
  4. Snoop sends an encrypted payment to Cozomo
  5. Cozomo sends his updated encrypted state to Snoop
  6. Snoop decrypts Cozomo’s new state, confirming the post-transaction balance (and that Cozomo actually got the $$ he was promised)

This elaborate dance has serious drawbacks: it’s expensive, it’s time consuming, and you can only dance with one person at a time — both parties have to be online at the same time to facilitate.

Worst of all, at the end of this dual-sided dialogue, neither party has convinced the rest of the world of anything — they’ve only mutually validated their one transaction.

Non bene.

Ain’t Note Fun?

But hol’ up — what if we flipped the attribution structure on its head? Ethereum defaults to an account model where an account has a balance. In other words, look up the account, and you get the balance.

What if we instead structured it to say a certain amount of money — described by a note — HAS an owner? Look up the note, and see who it belongs to.

Account has balance → note has owner

This is how Bitcoin works and it’s called UTXO (unspent transaction output). But forget the terminology. Think of UTXO’s as cash (bank notes).

Let’s think for a second about why cash is inherently more secure and private — or more precisely, more secure and private than account-based systems.

Cue the Jeopardy music:

Not the answer.

Got it? It’s secure because only the two parties transacting the cash know that ownership has changed hands! Everyone else in the entire universe can be kept in the dark.

You can think of a cash transaction as a change in ownership of an object (the note), whereas an accounting transaction is a change in the state of two accounts.

What an ownership change looks like for an encrypted note, probably.

When an Aztec transaction processes, rather than doing an account balance update (incrementing and decrementing balance), the network simply re-assigns ownership for a given note.

Why is this helpful? Well it turns out it’s way easier to encrypt a note, because it really only needs two things written on it: how much it’s worth, and who it’s owned by. When it changes hands, you scribble out the old owner’s name and write the new owner’s name. Ecco qua!

Simple Transfers on Aztec

So what exactly happens in a simple note transaction?

Say Snoop has two 50 ETH notes totaling 100 ETH and Cozomo has 0 notes.

Snoop’s two 50 ETH notes need to be destroyed, and two new notes created: an 80 ETH note that stays with Snoop, and a 20 ETH note that goes on to its new owner, Cozomo.

But how can privacy be preserved if the values of the notes have to be revealed?

Well — they don’t! Not publicly at least. Of course, Snoop and Cozomo know the value of their transaction, just like an exchange of cash, but they don’t have to reveal it to the world.

To protect their mutual privacy, Snoop publishes the transaction with a lock that he knows only Cozomo can unlock with his private key. The analogy here is kind of like putting the note in a little lock box. Of course, they both know what’s in the box (20 ETH), so Snoop has to trust Cozomo not to shout from the rooftop, “Someone just sent me 20 ETH!”

But otherwise, the note that was assigned new ownership goes back into a data structure holding all the notes that were ever created — a Merkle Tree hash, which we’ll cover in brief below.

What the state of the system looks like to an observer — the values and owners of each note fully encrypted.

Good Housekeeping

We know that Snoop destroyed two notes, created two new ones, and then sent one of the two new notes to his friend Cozomo. How can we make sure the two of them don’t collude to, for instance, double-spend? What if Snoop destroyed two notes worth 100 ETH in total, and created two new notes worth 200 ETH in total? Or, hell, an arbitrarily large amount?

Recall the two steps:

  1. Snoop destroys two 50 ETH notes and creates a 20 ETH note and an 80 ETH note
  2. Snoop sends the 20 ETH note to Cozomo

To ensure nothing fishy happens in step 1, all Snoop needs to do is prove to the system (Aztec) that the two notes he intends to create are equivalent in value to the two notes he intends to destroy.

This is known as a join-split transaction, and it conforms to this simple equivalence: A + B = C + D.

Here comes the psycho part. Buckle up.

In order to prove that the output notes (C + D) are equivalent in value to the input notes (A + B, or 100 ETH), Snoop generates a zero-knowledge proof (ZKP) locally, in his browser.

The black magic of ZKPs¹ means he can prove the equivalence A + B = C + D without revealing any of their individual values.

Aztec then validates the proof and says, “By the powers of Zero Knowledge, this must be true,” at which point the smart contract destroys the two input notes, generates the two output notes, and records the new output notes as an encrypted commitment in the note registry.

Proving Ownership

It’s worth discussing how ownership of notes is proven in Aztec, which has analogies to Ethereum-world. How do you prove you control access to an address in Ethereum? You sign a message using your wallet.

How do you prove you control access to a note in Aztec? With a very very fancy cryptographic signature called a zero knowledge proof.

The proof says, “Somewhere in Aztec’s state, there: a) exists a note with a certain value, and b) I own it.”

And how’s the state of the Aztec system stored? In two Merkle Trees:

  • A note tree, containing all the notes that have ever been created; and
  • A nullifier tree, containing all the notes that have ever been destroyed

Saying you own a note indicates to Aztec that the note exists in the note tree, and that no corresponding note-nullifier exists in the nullifier tree.

The extremely happy and normal note tree and the brooding, emo nullifier tree. If the nullifier tree could talk it would probably say something like “I f-ing hate you!” It’s okay honey, it’s okay. I know. It’s hard being 15.

When we talk about “destroying” a note, that actually means adding a nullifier to the nullifier tree rather than deleting a note from the note tree.

A humble Merkle Tree.

In order to send notes that I’ve proven I own, an entirely new Merkle tree (and Merkle root) is created. Once the Merkle roots of both the note tree and nullifier tree have moved to new values — in other words, the state of the system has been updated — those roots are published (settled) on Ethereum’s main chain and the transactions are deemed recorded.

Face Down, Bottom’s Up

I hope this gives you a solid grounding for understanding why privacy is challenging: we need to verify that transactions are legitimate and properly executed without violating or exposing user data.

These unique constraints mean privacy-preserving architecture must be constructed from the ground up. Aztec is the only L2 built this way — with privacy protected by its core architecture and facilitated by the magic of zero knowledge. Grazie mille!

Join the Aztec community

We’re always on the lookout for talented engineers and applied cryptographers. If joining our mission to bring scalable privacy to Ethereum excites you — get in touch with us at hello@aztecprotocol.com.

And continue the conversation with us on Discord or Twitter.

Want to a basic primer on zero-knowledge proofs? Check out this helpful YouTube video, this illustrated primer or Packy McCormick’s piece on the magic of ZK’s.

Fully Confidential Ethereum Transactions: Aztec Network’s Privacy Architecture was originally published in Aztec on Medium, where people are continuing the conversation by highlighting and responding to this story.

Read more
Aztec Network
Aztec Network
31 Mar
xx min read

Announcing the Alpha Network

Alpha is live: a fully feature-complete, privacy-first network. The infrastructure is in place, privacy is native to the protocol, and developers can now build truly private applications. 

Nine years ago, we set out to redesign blockchain for privacy. The goal: create a system institutions can adopt while giving users true control of their digital lives. Privacy band-aids are coming to Ethereum (someday), but it’s clear we need privacy now, and there’s an arms race underway to build it. Privacy is complex, it’s not a feature you can bolt-on as an afterthought. It demands a ground-up approach, deep tech stack integration, and complete decentralization.

In November 2025, the Aztec Ignition Chain went live as the first decentralized L2 on Ethereum, it’s the coordination layer that the execution layer sits on top of. The network is not operated by the Aztec Labs or the Aztec Foundation, it’s run by the community, making it the true backbone of Aztec. 

With the infrastructure in place and a unanimous community vote, the network enters Alpha. 

What is the Alpha Network?

Alpha is the first Layer 2 with a full execution environment for private smart contracts. All accounts, transactions, and the execution itself can be completely private. Developers can now choose what they want public and what they want to keep private while building with the three privacy pillars we have in place across data, identity, and compute.

These privacy pillars, which can be used individually or combined, break down into three core layers: 

  1. Data: The data you hold or send remains private, enabling use cases such as private transactions, RWAs, payments and stablecoins.
  2. Identity: Your identity remains private, enabling accounts that privately connect real world identities onchain, institutional compliance, or financial reporting where users selectively disclose information.
  3. Compute: The actions you take remain private, enabling applications in private finance, gaming, and beyond.

The Key Components  

Alpha is feature complete–meaning this is the only full-stack solution for adding privacy to your business or application. You build, and Aztec handles the cryptography under the hood. 

It’s Composable. Private-preserving contracts are not isolated; they can talk to each other and seamlessly blend both private and public state across contracts. Privacy can be preserved across contract calls for full callstack privacy. 

No backdoor access. Aztec is the only decentralized L2, and is launching as a fully decentralized rollup with a Layer 1 escape hatch.

It’s Compliant. Companies are missing out on the benefits of blockchains because transparent chains expose user data, while private networks protect it, but still offer fully customizable controls. Now they can build compliant apps that move value around the world instantly.

How Apps Work on Alpha 

  1. Write in Noir, an open-source Rust-like programming language for writing smart contracts. Build contracts with Aztec.nr and mark functions private or public.
  1. Prove on a device. Users execute private logic locally and a ZK proof is generated.
  1. Submit to Aztec. The proof goes to sequencers who validate without seeing the data. Any public aspects are then executed.
  1. Settle on Ethereum. Proofs of transactions on Aztec are settled to Ethereum L1.

Developers can explore our privacy primitives across data, identity, and compute and start building with them using the documentation here. Note that this is an early version of the network with known vulnerabilities, see this post for details. While this is the first iteration of the network, there will be several upgrades that secure and harden the network on our path to Beta. If you’d like to learn more about how you can integrate privacy into your project, reach out here

To hear directly from our Cofounders, join our live from Cannes Q&A on Tuesday, March 31st at 9:30 am ET. Follow us on X to get the latest updates from the Aztec Network.

Aztec Network
Aztec Network
27 Mar
xx min read

Critical Vulnerability in Alpha v4

On Wednesday 17 March 2026 our team discovered a new vulnerability in the Aztec Network. Following the analysis, the vulnerability has been confirmed as a critical vulnerability in accordance with our vulnerability matrix.

The vulnerability affects the proving system as a whole, and is not mitigated via public re-execution by the committee of validators. Exploitation can lead to severe disruption of the protocol and theft of user funds.

In accordance with our policy, fixes for the network will be packaged and distributed with the “v5” release of the network, currently planned for July 2026.

The actual bug and corresponding patch will not be publicly disclosed until “v5.”

Aztec applications and portals bridging assets from Layer 1s should warn users about the security guarantees of Alpha, in particular, reminding users not to put in funds they are not willing to lose. Portals or applications may add additional security measures or training wheels specific to their application or use case.

State of Alpha security

We will shortly establish a bug tracker to show the number and severity of bugs known to us in v4. The tracker will be updated as audits and security researchers discover issues. Each new alpha release will get its own tracker. This will allow developers and users to judge for themselves how they are willing to use the network, and we will use the tracker as a primary determinant for whether the network is ready for a "Beta" label.

Additional bug disclosure

We have identified a vulnerability in barretenberg allowing inclusion of incorrect proofs in the Aztec Network mempool, and ask all nodes to upgrade to versions v.4.1.2 or later.

We’d like to thank Consensys Diligence & TU Vienna for a recent discovery of a separate vulnerability in barretenberg categorized as medium for the network and critical for Noir:

We have published a fixed version of barretenberg.

We’d also like to thank Plainshift AI for discovery, reproduction, and reporting of one more vulnerability in the Aztec Network and their ongoing work to help secure the network.

Aztec Network
Aztec Network
18 Mar
xx min read

How Aztec Governance Works

Decentralization is not just a technical property of the Aztec Network, it is the governing principle. 

No single team, company, or individual controls how the network evolves. Upgrades are proposed in public, debated in the open, and approved by the people running the network. Decentralized sequencing, proving, and governance are hard-coded into the base protocol so that no central actor can unilaterally change the rules, censor transactions, or appropriate user value.

The governance framework that makes this possible has three moving parts: Aztec Improvement Proposal (AZIP), Aztec Upgrade Proposal (AZUP), and the onchain vote. Together, they form a pipeline that takes an idea to a live protocol change, with multiple independent checkpoints along the way.

The Virtual Town Square

Every upgrade starts with an AZIP. AZIPs are version-controlled design documents, publicly maintained on GitHub, modeled on the same EIP process that has governed Ethereum since its earliest days. Anyone is encouraged to suggest improvements to the Aztec Network protocol spec.

Before a formal proposal is opened, ideas live in GitHub Discussions, an open forum where the community can weigh in, challenge assumptions, and shape the direction of a proposal before it hardens into a spec. This is the virtual town square: the place where the network's future gets debated in public, not decided behind closed doors.

The AZIP framework is what decentralization looks like in practice. Multiple ideas can surface simultaneously, get stress-tested by the community, and the strongest ones naturally rise. Good arguments win, not titles or seniority. The process selects for quality discussion precisely because anyone can participate and everything is visible.

Once an AZIP is formalized as a pull request, it enters a structured lifecycle: Draft, Ready for Discussion, then Accepted or Rejected. Rejected AZIPs are not deleted — they remain permanently in the repository as a record of what was tried and why it was rejected. Nothing gets quietly buried.

Security Considerations are mandatory for all Core, Standard, and Economics AZIPs. Proposals without them cannot pass the Draft stage. Security is structural, not an afterthought.

From Proposal to Upgrade

Once Core Contributors, a merit-based and informal group of active protocol contributors, have reviewed an AZIP and approved it for inclusion, it gets bundled into an AZUP.

An AZUP takes everything an AZIP described and deploys it — a real smart contract, real onchain actions. Each AZUP includes a payload that encodes the exact onchain changes that will occur if the upgrade is approved. Anyone can inspect the payload on a block explorer and see precisely what will change before voting begins.

The payload then goes to sequencers for signaling. Sequencers are the backbone of the network. They propose blocks, attest to state, and serve as the first governance gate for any upgrade. A payload must accumulate enough signals from sequencers within a fixed round to advance. The people actually running the network have to express coordinated support before any change reaches a broader vote.

Once sequencers signal quorum, the proposal moves to tokenholders. Sequencers' staked voting power defaults to "yea" on proposals that came through the signaling path, meaning opposition must be active, not passive. Any sequencer or tokenholder who wants to vote against a proposal must explicitly re-delegate their stake before the voting snapshot is taken. The system rewards genuine engagement from all sides.

For a proposal to pass, it must meet quorum, a supermajority margin, and a minimum participation threshold, all three. If any condition is unmet, the proposal fails.

Built-In Delays, Built-In Safety

Even after a proposal passes, it does not execute immediately. A mandatory delay gives node operators time to deploy updated software, allows the community to perform final checks, and reduces the risk of sudden uncoordinated changes hitting the network. If the proposal is not executed within its grace period, it expires.

Failed AZUPs cannot be resubmitted. A new proposal must be created that directly addresses the feedback received. There is no way to simply retry and hope for a different result.

No Single Point of Control

The teams building the network have no special governance power. Sequencers, tokenholders, and Core Contributors are the governing actors, each playing a distinct and non-redundant role.

No single party can force or block an upgrade. Sequencers can withhold signals. Tokenholders can vote nay. Proposals not executed within the grace period expire on their own.

This is decentralization working as intended. The network upgrades not because a team decides it should, but because the people running it agree that it should.

If you want to help shape what Aztec becomes, the forum is open. The proposals are public. The town square is yours. 

Follow Aztec on X to stay up to date on the latest developments.

Aztec Network
Aztec Network
10 Mar
xx min read

Alpha Network Security: What to Expect

Aztec’s Approach to Security

Aztec is novel code — the bleeding edge of cryptography and blockchain technology. As the first decentralized L2 on Ethereum, Aztec is powered by a global network of sequencers and provers. Decentralization introduces some novel challenges in how security is addressed; there is no centralized sequencer to pause or a centralized entity who has power over the network. The rollout of the network reflects this, with distinct goals at each phase.

Ignition

Validate governance and decentralized block building work as intended on Ethereum Mainnet. 

Alpha

Enable transactions at 1TPS, ~6s block times and improve the security of the network via continual ongoing audits and bug bounty. New releases of the alpha network are expected regularly to address any security vulnerabilities. Please note, every alpha deployment is distinct and state is not migrated between Alpha releases. 

Beta

We will transition to Beta once the network scales to >10 TPS, with reduced block times while ensuring 99.9% uptime. Additionally, the transition requires no critical bugs disclosed via bug bounty in 3 months. State migrations across network releases can be considered.

TL;DR: The roadmap from Ignition to Alpha to Beta is designed to reflect the core team's growing confidence in the network's security.

This phased approach lets us balance ecosystem growth while building security confidence and steadily expanding the community of researchers and tools working to validate the network’s security, soundness and correctness.

Ultimately, time in production without an exploit is the most reliable indicator of how secure a codebase is.

At the start of Alpha, that confidence is still developing. The core team believes the network is secure enough to support early ecosystem use cases and handle small amounts of value. However this is experimental alpha software and users should not deposit more value than they are willing to lose. Apps may choose to limit deposit amounts to mitigate risk for users.

Audits are ongoing throughout Alpha, with the goal to achieve dual external audits across the entire codebase.

The table below shows current security and audit coverage at the time of writing.

The main bug bounty for the network is not yet live, other than for the non-cryptographic L1 smart contracts as audits are ongoing. We encourage security researchers to responsibly disclose findings in line with our security policy .

As the audits are still ongoing, we expect to discover vulnerabilities in various components. The fixes will be packaged and distributed with the “v5” release.

If we discover a Critical vulnerability in “v4” in accordance with the following severity matrix, which would require the change of verification keys to fix, we will first alert the portal operators to pause deposits and then post a message on the forum, stating that the rollup has a vulnerability.

Security of the Aztec Virtual Machine (AVM)

Aztec uses a hybrid execution model, handling private and public execution separately — and the security considerations differ between them.

As per the audit table above, it is clear that the Aztec Virtual Machine (AVM) has not yet completed its internal and external audits. This is intentional as all AVM execution is public, which allows it to benefit from a “Training Wheel” — the validator re-execution committee.

Every 72 seconds, a collection of newly proposed Aztec blocks are bundled into a "checkpoint" and submitted to L1. With each proposed checkpoint, a committee of 48 staking validators randomly selected from the entire set of validators (presently 3,959) re-execute all txs of all blocks in the checkpoint, and attest to the resulting state roots. 33 out of 48 attestations are required for the checkpoint proposal to be considered valid. The committee and the eventual zk proof must agree on the resultant state root for a checkpoint to be added to the proven chain. As a result, an attacker must control 33/48 of any given committee to exploit any bug in the AVM.

The only time the re-execution committee is not active is during the escape hatch, where the cost to propose a block is set at a level which attempts to quantify the security of the execution training wheel. For this version of the alpha network, this is set a 332M AZTEC, a figure intended to approximate the economic protection the committee normally provides, equivalent to roughly 19% of the un-staked circulating supply at the time of writing. Since the Aztec Foundation holds a significant portion of that supply, the effective threshold is considerably higher in practice.

Quantifying the cost of committee takeover attacks

A key design assumption is that just-in-time bribery of the sequencer committee is impractical and the only ****realistic attack vector is stake acquisition, not bribery.

Assuming a sequencer set size of 4,000 and a committee that rotates each epoch (~38.4mins) from the full sequencer set using a Fisher-Yates shuffle seeded by L1 RANDAO we can see the probability and amount of stake required in the table below.

To achieve a 99% probability of controlling at least one supermajority within 3 days, an attacker would need to control approximately 55.4% of the validator set - roughly 2,215 sequencers representing 443M AZTEC in stake. Assuming an exploit is successful their stake would likely de-value by 70-80%, resulting in an expected economic loss of approximately 332M AZTEC.

To achieve only a 0.5% probability of controlling at least one supermajority within 6 months, an attacker would need to control approximately 33.88% of the validator set.

What does this means for builders?

The practical effect of this training wheel is that the network can exist while there are known security issues with the AVM, as long as the value an attacker would gain from any potential exploit is less than the cost of acquiring 332M AZTEC.

The training wheel allows security researchers to spend more time on the private execution paths that don’t benefit from the training wheel and for the network to be deployed in an alpha version where security researchers can attempt to find additional AVM exploits.

In concrete terms, the training wheel means the Alpha network can reasonably secure value up to around 332M AZTEC (~$6.5M at the time of writing).

Ecosystem builders should keep the above limits in mind, particularly when designing portal contracts that bridge funds into the network.

Portals are the main way value will be bridged into the alpha network, and as a result are also the main target for any exploits. The design of portals can allow the network to secure far higher value. If a portal secures > 332M AZTEC and allows all of its funds to be taken in one withdrawal without any rate limits, delays or pause functionality then it is a target for an AVM exploit attack.

If a portal implements a maximum withdrawal per user, pause functionality or delays for larger withdrawals it becomes harder for an attacker to steal a large quantum of funds in one go.

Conclusion

The Aztec Alpha code is ready to go. The next step is for someone in the community to submit a governance proposal and for the network to vote on enabling transactions. This is decentralization working as intended.

Once live, Alpha will run at 1 TPS with roughly 6 second block times. Audits are still ongoing across several components, so keep deposits small and only put in what you're comfortable losing.

On the security side, a 48-validator re-execution committee provides the main protection during Alpha, requiring 33/48 consensus on every 72-second checkpoint. Successfully attacking the AVM would require controlling roughly 55% of the validator set at a cost of around 332M AZTEC, putting the practical security ceiling at approximately $6.5M.

Alpha is about growing the ecosystem, expanding the security of the network, and accumulating the one thing no audit can shortcut: time in production. This is the network maturing in exactly the way it was designed to as it progresses toward Beta.