# Aztec discloses a proving flaw in its Alpha V5 network that could let forged transactions pass verification

Core contributors found the bug with AI-assisted auditing on July 27, kept it private until a fix was ready for V6, and say chain history cannot show whether anyone used it.

- Published: 2026-10-14T05:36:59.411Z
- Canonical: https://polylog.news/crypto/2026-10-14/aztec-discloses-a-proving-flaw-in-its-alpha-v5-network-that
- 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 Network](https://aztec.network/blog/how-gas-works-on-aztec)

Aztec, the privacy-focused rollup on Ethereum, has disclosed [a critical soundness vulnerability in the proving system of its V5 Alpha network](https://aztec.network/blog/alpha-v5-proving-system-vulnerability). A soundness flaw means the system can accept a proof of something false. According to Aztec, an attacker could build a proof that passes verification for a transaction the network should reject. That transaction could then produce a state change outside the rules V5 is meant to enforce.

Core contributors say they found the flaw on July 27, 2026, through internal AI-assisted auditing. They delayed public disclosure until a fix was scheduled for [the next release, V6](https://aztec.network/alpha-v5). V6 will include circuit updates that stop the network from accepting proofs tied to the affected proving system. Aztec says reviewers have found no other high-severity or critical V5 issues so far.

The most important admission concerns the past. Aztec states that it cannot determine whether anyone exploited the flaw before it was found. The affected system does not record enough information to tell a normally accepted transaction from one accepted through the faulty proving path. Historical chain data therefore cannot show whether exploitation occurred or how large any damage was. Contributors are working with node operators, application teams and bridge operators, who are adding security guards around affected flows until V6 ships.

This is not Aztec's first such disclosure. The team previously published [a critical vulnerability in Alpha v4](https://aztec.network/blog/critical-vulnerability-in-alpha-v4), and it [raised its total bug bounty to $2 million](https://aztec.network/blog/aztec-network-raises-total-bug-bounty-to-2-million). In a privacy rollup, the network hides transaction contents by design. Other validators cannot re-execute private transactions to check them, so the proof is the only thing standing between users and an invalid state change. Aztec has also [published how fees and gas work on the network](https://aztec.network/blog/how-gas-works-on-aztec) as it moves toward a token-backed mainnet. A soundness bug there would affect real economic value rather than test balances.

## What this means

The mechanism is specific to zero-knowledge systems. When a proof is the only check, a soundness bug can forge state without leaving a trace, and Aztec confirms that its own records cannot reconstruct what happened. The exposed parties are bridge operators and applications holding value on Aztec, and more broadly every rollup whose security depends on one proving system. Markets treat rollup risk mainly as smart-contract risk. This case shows that a separate class of risk can be invisible after the fact, and AI tools are now finding it. The desk's read is that, over the coming months, more rollups will adopt multiple independent provers and bridge-side withdrawal limits. Disclosures timed to coincide with upgrades will become the normal practice. One condition matters: a second major proving-system bug must surface at another team. The trend weakens if V6 ships and outside auditors of the new circuits find nothing over several months.

## What to watch

- The V6 release date and whether independent auditors publish reviews of the new circuits, because that decides when bridges can lift their temporary guards.
- Whether any bridge or application on Aztec reports a balance inconsistency, which would be the only indirect evidence that the flaw was used.
- Whether other zero-knowledge rollups disclose similar AI-found soundness bugs, which would confirm that AI review is clearing a backlog across the sector.