# Aztec Says a Flaw in Its Alpha V5 Proving System Could Let an Invalid Transaction Pass Verification

Core contributors found the defect on 27 July through internal artificial-intelligence-assisted auditing, after internal and external human audits had already finished without catching it.

- Published: 2026-09-11T05:49:36.633Z
- Canonical: https://polylog.news/crypto/2026-09-11/aztec-says-a-flaw-in-its-alpha-v5-proving-system-could-let-a
- Publisher: Polylog (Crypto desk)
- Section: crypto
- Sources: [Aztec Network](https://aztec.network/blog/alpha-v5-proving-system-vulnerability), [Aztec Alpha V5](https://aztec.network/alpha-v5)

Aztec, the privacy-focused Ethereum layer-2 network, disclosed a critical defect in the proving system behind its [Alpha V5 release](https://aztec.network/alpha-v5). According to the team, [an attacker may be able to construct a proof that passes verification for a transaction the network should reject](https://aztec.network/blog/alpha-v5-proving-system-vulnerability). That is a soundness failure, not a liveness problem: the verifier wrongly accepts an invalid transaction as valid.

Two details matter more than the bug itself. Core contributors identified it on 27 July 2026 through internal auditing assisted by artificial intelligence, after internal and external human audits had already completed. And the team states it cannot determine whether anyone exploited the flaw before the finding. Aztec says funds, applications and contract state on V5 should be treated as exposed to a protocol-level failure until operators complete the required network actions, and that reviewers have found no other critical or high-severity issues in V5 so far.

Soundness bugs sit in a different risk class from smart-contract exploits. A rollup that relies on validity proofs has no fallback of re-execution by independent validators to catch a forged proof, so the proving stack becomes a single point of correctness for every user balance it secures. Layer-2 networks tracked by L2Beat currently secure $48.85 billion in total value.

## What this means

The failure surface for zero-knowledge rollups is the circuit and the verifier, not the application code that auditors usually price. Users and applications on proof-based networks are exposed through a channel they cannot inspect, because a forged proof looks identical to a valid one on chain. The disclosure also carries a second signal for the audit industry: machine-assisted review caught what completed human audits missed, which shifts where security budgets go. Two outcomes are in play. Either proof-system diversity and working escape hatches become standard requirements before networks hold serious value, or teams keep launching single-prover systems and the next such bug is found by an attacker rather than an auditor.

## What to watch

- Whether Aztec publishes the technical root cause and the affected circuit after operators upgrade, which is the test of how seriously the ecosystem treats embargoed proving-system disclosure.
- Whether other zero-knowledge rollups run the same class of check against their own verifiers and report results, or stay silent.
- Whether machine-assisted auditing starts finding defects in production proving systems, not only in alpha releases, which would change how much value teams are willing to secure with a single prover.