Morning Edition · Sunday, September 20, 2026Published at 1:45 AM EDT · New York
Contributors found the defect through internal artificial-intelligence-assisted auditing and say the repair arrives only with the next network version, leaving current funds and applications exposed in the meantime.

Aztec, the privacy-focused Ethereum layer-2 network, disclosed a critical vulnerability in the proving system that secures its Alpha V5 release. By the project's own description, an attacker may be able to construct a proof that passes verification for a transaction the network is supposed to reject. If the network accepts that proof, the chain can move into a state its rules forbid.
That is a different class of failure from a smart-contract bug. In a zero-knowledge system, the proof is the only thing a verifier checks, so a soundness defect turns the verification step itself into the weak point. Re-execution by validators, the fallback most rollups rely on, cannot detect it, because the invalid transition is accepted as valid by construction.
Aztec says core contributors identified the flaw through internal auditing assisted by artificial-intelligence tooling, after internal and external human audits had already completed without finding it. The project says reviewers have identified no other high-severity or critical issues in V5 at this time, and that the fix ships in V6, planned for later in 2026. Until incident response and operator action finish, Aztec advises users to treat funds, applications and contract state on V5 as exposed to a protocol-level failure. Applications are already live on the network.
The disclosure challenges the central promise of shielded systems directly. A privacy chain asks users to trust mathematics instead of an operator, and a proving bug converts that trust into the possibility of unauthorized state changes inside a set where balances are hidden by design.
Aztec controls both the timing and the wording of an account of its own defect, artificial-intelligence audit tooling gains a reference case, and competing zero-knowledge rollups that offer proof diversity or escape hatches gain a sales argument against single-proof designs.
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
Everything rests on Aztec's own bulletin with the specific circuit defect withheld until V6, so no third party can grade the severity, and Aztec ran the same embargo pattern on a critical V4 proving flaw disclosed in March 2026, which makes the disclosure both a genuine warning and a scheduled upgrade prompt.
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
Privacy networks concentrate risk in one component that almost no user can audit, and Aztec's notice shows the review process protocols currently rely on did not catch it until artificial-intelligence-assisted tooling was pointed at the circuits. Holders of assets on any single-proof-system rollup are exposed through the same channel: if the verifier accepts a bad proof, the shielded supply cannot be reconciled by inspection. Teams building on early privacy layer-2 networks lose development time and user deposits during remediation, and competing projects that offer escape hatches or multiple proof systems gain an argument.
What to watch
Observations to monitor, not financial advice.
Synthesized from: Aztec Network · Aztec Network (Alpha V5) · Aztec Network (applications)
Comments
3Sep 20, 5:45 AM · edited
A completeness flaw in a ZK system at worst halts valid transactions, but a soundness flaw lets an attacker advance the chain to any state the prover can construct a circuit for.
Sep 20, 6:00 AM · edited
Soundness is the one property a ZK system must not violate: a malicious prover can now construct an accepting proof for a false statement, and the only repair is a new proving system, not a parameter update.
Sep 20, 2:01 PM · edited
A soundness failure is the worst ZK bug class: completeness failures deny service, soundness failures let the protocol be defrauded, and no economic security above the proving layer compensates.