# Aztec Discloses Critical Proving-System Flaw in Its Alpha V5 Privacy Network

Contributors say an attacker could construct a proof that verifies for a transaction the network should reject, and the fix is deferred to the next release.

- Published: 2026-09-29T05:41:12.286Z
- Canonical: https://polylog.news/crypto/2026-09-29/aztec-discloses-critical-proving-system-flaw-in-its-alpha-v5
- Publisher: Polylog (Crypto desk)
- Section: crypto
- Sources: [Aztec Network](https://aztec.network/blog/alpha-v5-proving-system-vulnerability), [Aztec Network](https://aztec.network/alpha-v5)

Aztec, the privacy-focused layer-2 network on Ethereum, has disclosed a critical vulnerability in the proving system behind its [Alpha V5 release](https://aztec.network/alpha-v5). Core contributors say the defect could let an attacker build a proof that passes verification for a transaction the network should have rejected, and that funds, applications and contract state on V5 should be treated as exposed to a protocol-level failure until operators complete the required network actions. The [disclosure](https://aztec.network/blog/alpha-v5-proving-system-vulnerability) states that the fix is planned for the V6 release and that the finding came out of internal auditing assisted by artificial intelligence tools, after internal and external human audits had already been completed.

This is a soundness bug, a different class of problem from the smart-contract exploits that dominate loss tables. A soundness defect does not require a buggy application. It breaks the guarantee that the verifier only accepts statements that are true, an assumption every contract on the chain inherits. Aztec has been through this before. It [told V4 users to withdraw](https://thedefiant.io/news/defi/aztec-v4-withdraw-june-25-v5-upgrade-security-vulnerability) ahead of the V5 upgrade after an earlier critical finding.

The structural problem for privacy chains is that the usual fallback does not apply. Optimistic rollups and many validity rollups rely on node re-execution or fraud windows as a backstop. On a network where execution happens on the user's own device and the chain sees only a commitment, no independent party can re-run the transaction and notice that the arithmetic was wrong. The proof is the only check.

Aztec's framing is that Alpha exists precisely to surface findings like this before wider deployment, a defensible position for a network still labeled alpha. The counterpoint is that users who bridged real value into V5 carried the risk of an unproven verifier with no compensating mechanism, and that the disclosure and the fix are separated by a full release cycle.

## What this means

The exposure runs to anyone holding assets or state on Aztec V5 and, indirectly, to every team building on a single proving stack without a fallback safeguard. A verifier that can be persuaded to accept a false statement is a mint function for whoever finds it, which is why disclosure is usually timed to an upgrade rather than published when the bug is found. Aztec is choosing disclosure over embargo, which reduces the chance of a quiet exploit but leaves V5 exposed until V6 ships. The precedent worth naming is the series of soundness findings in major zero-knowledge stacks over the past two years, none of which were exploited in production but all of which pushed teams toward multi-prover designs and audited fallbacks. Expect the practical consequence within months to be more privacy and validity rollups announcing second provers, formal verification of circuit constraints, and explicit withdrawal windows. Expect capital to stay in the low hundreds of millions on alpha-stage privacy chains until those safeguards exist. The read is wrong if an attacker exploits the flaw before V6 ships, which would move the question from engineering practice to solvency.

## What to watch

- Whether Aztec sets a hard date for V6 and a forced withdrawal window for V5 balances, which would indicate how severe contributors consider the exposure.
- Whether other zero-knowledge teams respond with multi-prover commitments, since proof-system diversity is the only defense that does not depend on finding every bug first.
- How much value users bridge into Aztec after the disclosure, which shows whether privacy demand outweighs verifier risk for real capital.