Morning Edition · Wednesday, September 16, 2026Published at 1:44 AM EDT · New York
Core contributors found the defect through internal AI-assisted auditing and told users to treat funds, applications and contract state on the Alpha V5 network as exposed until operators complete the required network actions.

Aztec, the privacy-focused Ethereum layer-2 network, disclosed a critical vulnerability in the proving system behind its Alpha V5 release. According to the project's disclosure, an attacker could construct a proof that passes verification for a transaction the network should reject, producing a state transition outside the rules the protocol intends to enforce. Core contributors said they found the defect through internal auditing assisted by artificial intelligence (AI) tools. They advised users to treat funds, applications and contract state on V5 as exposed until incident response is complete and network operators carry out the required actions.
The failure class matters more than this particular network. In an optimistic rollup, validators re-execute transactions and can catch a bad state transition. In a zero-knowledge system, the proof is the only thing standing between a submitted transaction and accepted state, so a soundness defect in the proving stack is not caught by anyone re-running the computation. It is caught by auditing, formal verification, or by an attacker.
This is the second such episode for Aztec within months. When the project moved from V4 to V5, it told users to withdraw before June 25 because the upgrade had exposed critical flaws in the earlier release. Aztec has raised its bug bounty to $2 million and says more than 234 researchers have engaged with the program on Cantina. The network remains in alpha, which limits the value at risk, and the contributors said further bugs should be expected as usage grows.
The disclosure can be read two ways. Disclosing a soundness bug found in-house, before an attacker used it, is the behavior a serious engineering organization should show. At the same time, each disclosure confirms that privacy rollups are shipping systems whose security nobody, including their authors, can yet fully verify, and that users who bridge assets into them are accepting that residual risk in exchange for confidentiality.
Part of a tracked trend
Proving-System Bugs Become a Distinct Rollup Risk
Soundness defects in zero-knowledge proving systems will keep surfacing as a risk class separate from smart-contract exploits, because validator re-execution — the fallback most rollups rely on — cannot catch them, forcing teams into embargoed disclosure timed to upgrades and pushing users toward proof-system diversity and escape hatches.
Start a discussion in Townsquare.
More from this edition
Aztec gains reputational credit for self-reporting a soundness defect before anyone exploited it, and competing rollup teams gain a case study to argue for proof-system diversity.
The only source for the flaw, its severity and its discovery method is Aztec's own disclosure, no independent researcher has confirmed it, and the post dates the finding to 27 July 2026 rather than this week, so the timing implied by placing it in a September edition does not match the disclosure record.
An open-source-intelligence read of how likely this story is true with its real nuance, not a judgment of any outlet. It assesses the claim, weighing independent and adversarial reporting. How we label confidence.
What this means
Soundness bugs in zero-knowledge proving systems bypass the fallback that most rollup users implicitly rely on, because no validator re-executes the transaction to notice the state is wrong. That exposes anyone holding assets on a proof-only chain to a total-loss event rather than a partial one, and it pushes serious deployments toward proof-system diversity, escape hatches and delayed upgrade windows. AI-assisted auditing now serves both defenders and attackers. It found this defect before an attacker did, and it is the same capability attackers can use against published circuits.
What to watch
Observations to monitor, not financial advice.
Synthesized from: Aztec Network · Aztec Network (Alpha V5) · Aztec Network (protocol blog)
Comments
0No comments yet.