# Aztec Says a Flaw in Its Alpha V5 Proving System Can Make an Invalid Transaction Verify

Core contributors found the defect on July 27 through internal auditing assisted by artificial intelligence, and told users to treat funds, applications and contract state on V5 as exposed until operators complete the response.

- Published: 2026-08-20T05:47:26.932Z
- Canonical: https://polylog.news/crypto/2026-08-20/aztec-says-a-flaw-in-its-alpha-v5-proving-system-can-make-an
- 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/dark-forest-aztec-game-goes-live)

Aztec, the privacy-focused layer-2 network on Ethereum, [disclosed a critical vulnerability](https://aztec.network/blog/alpha-v5-proving-system-vulnerability) in the proving system behind its Alpha V5 release. The described failure is a soundness defect: an attacker may be able to construct a zero-knowledge proof that passes verification for a transaction the network should have rejected. Contributors identified the issue on July 27 through internal auditing assisted by artificial intelligence, and said funds, applications and contract state on V5 should be treated as exposed to a protocol-level failure until incident response finishes and operators carry out the required network actions.

This is a different risk class from a smart-contract bug. On a network that hides transaction contents, the proof is the only thing a verifier checks. There is no re-execution fallback, because no other party can see the inputs to recompute them. A soundness break therefore reaches the whole system at once rather than one application.

It is also the second such disclosure in the same programme. Aztec published a critical vulnerability in Alpha V4 earlier this year, and its governance process required the underlying V4 bugs to be made public when the [V5 upgrade](https://aztec.network/alpha-v5) went on-chain. The project continues to ship application-layer work, including [a live on-chain game built on private state](https://aztec.network/blog/dark-forest-aztec-game-goes-live), while the base layer remains explicitly labelled alpha software.

The straightforward reading is that the disclosure process worked as designed and that the technology is not finished. Both statements are true at the same time.

## What this means

Anyone holding assets or state on a zero-knowledge network is trusting the proving stack, not just the contracts audited on top of it. Aztec's disclosure shows that a single circuit or verifier defect can invalidate everything downstream, and that privacy removes the re-execution check other rollups rely on. Teams building confidential settlement for institutions are the ones most exposed, because a soundness bug is exactly the failure an institutional risk committee cannot underwrite. The offsetting fact is that these bugs are surfacing during alpha testing, found internally, rather than being found first by an attacker.

## What to watch

- Whether Aztec publishes the technical write-up of the specific circuit defect after the upgrade, which is what other proving teams need in order to check their own systems for the same pattern.
- Whether other zero-knowledge rollups add proof-system diversity or independent verifiers, since that is the only structural answer to a single-proof failure.
- How much of the artificial-intelligence-assisted audit work by Aztec and other teams finds further soundness bugs, which would indicate the backlog of undiscovered defects is large rather than nearly cleared.