Every privacy system has an off-switch. What matters is who holds it, and on most networks the answer is an operator. Here is why Aztec was built so that nobody does.
Every private system can be opened. A court can order it, an auditor can ask for it, a user can choose to show their books. That is fine, and often necessary. The question that separates privacy networks from each other is who gets to do the opening.
On a growing number of networks, the answer is someone other than you.
The fastest way to offer private transactions is to route them through a party that can read them. The details vary. One network has a zone operator who sees every transaction in plaintext. Another runs a private shard with an approved list of validators and trusted hardware. Another encrypts balances but keeps the decryption key with a committee of thirteen operators. Another stores each user's viewing key, encrypted, where a designated auditing company can recover it.
These designs ship quickly and they feel familiar, because they work the way a bank does. There is a responsible party in the middle, and you trust them.
The trouble is what happens to that party. Zac Williamson put it plainly in The Devil's Bargain: "If one party can see everything and stop anything, that party may be treated as responsible for seeing and stopping." An operator with the ability to read transactions will be asked to read them. An operator with the ability to block a user will be told to block one. None of that needs bad intent. A subpoena will do it. So will a breached server or a change of owner.
Privacy that exists at an operator's discretion lasts until the day the operator's incentives change.
The Ethereum community has a term for the alternative: credible neutrality. A credibly neutral system is one you can check for yourself, without taking the builder's word, and find that it does not favour or target anyone. For a privacy network that comes down to two questions.
Can anyone read my data without me? And can anyone stop me from using the network?
If the honest answer to either is "yes, but they promise not to", the network has an off-switch and someone else is holding it.
On Aztec, the private part of a transaction never leaves your device. Your wallet runs the contract's private functions locally, against encrypted notes only you can decrypt, and produces a zero-knowledge proof that the rules were followed. The network receives the proof, some commitments, and some nullifiers. It checks the maths. It does not see who paid whom or how much, because that information was never sent.
There is no operator copy of your viewing key, no auditor escrow, no enclave you have to believe in, and no committee that could decrypt the chain if enough of its members agreed to.
Disclosure still works. It just runs in the other direction. When you need to prove something to an auditor or a counterparty, you hand over the data yourself, together with a proof that it matches what is on chain. You choose the fact and you choose the recipient. Nobody can make that choice for you, and nobody can make it behind your back.
The second test is about who runs the thing. Aztec's Ignition Chain went live in November 2025 with no company in the operating seat. More than 3,500 sequencers take turns proposing blocks, selected at random from everyone who has staked. More than 50 provers compete to prove them. Anyone who meets the published requirements can join either group, and neither Aztec Labs nor the Aztec Foundation runs the network.
L2BEAT, which has no reason to flatter anyone, rates Aztec Stage 2, its highest category. Users can exit at any time because the contracts cannot be upgraded out from under them.
Changing the protocol takes three separate groups. Sequencers have to signal support for a proposal while they produce blocks. Tokenholders then vote. A timelock runs before anything executes, so node operators and users can see what is coming. No one of those groups can push a change through alone, and no one of them can quietly block one.
This design is slower to build and it asks more of the user's device. Proving a private transfer takes about 2.5 seconds on a laptop in Alpha V5, and closer to 7 in a browser. Sending plaintext to a server would be instant. Governance by vote and timelock is slower than a company deciding on a Tuesday. And the network is Alpha software: it is live on Ethereum mainnet, it is being hardened in public, and nobody should treat it as finished.
We think those are the right costs. Proving time halved between V4 and V5, and that work continues. The property you get in return does not erode. A network that cannot read your data cannot be compelled to hand it over, and a network with no operator has nobody to lean on.
You do not need to read a whitepaper to evaluate a privacy network. Ask who could read your transactions if they wanted to. Then ask who could stop you from making one.
On Aztec the answer to both is nobody. We built it that way on purpose, and you can check.
Introducing the Aztec Foundation, a nonprofit organization to support the growth and development of open-source programmable privacy. The launch of the Aztec Foundation marks a significant milestone for the Aztec Network, bringing us closer to the launch of a fully decentralized, privacy-preserving network.
Ariel Gabizon, a distinguished cryptographer, coauthor of PLONK, and pioneer of zero-knowledge (ZK) technology, has rejoined Aztec Labs as its Chief Scientist. With Devnet now live and private, client-side smart contract execution with public verifiability possible, Ariel is rejoining us at a pivotal moment for Aztec, our upcoming privacy-first L2 on Ethereum.
This piece, written by Zac Williamson with insights from Arnaud Schenk, delves into the evolving landscape of blockchain technology and its impact on societal structures.