Aztec Foundation co-founder and CEO Zac Williamson has acknowledged the possibility that more advanced AI models could uncover bugs in Aztec Tierra (Alpha V6), the latest version of the Ethereum privacy network, which launched on Nov. 3. In an interview recorded on Nov. 9 for Unchained Premium, Williamson stated that while AI audits on the previous version identified some critical flaws in its cryptographic libraries, the team cannot rule out future discoveries by more sophisticated AI.
Williamson highlighted that AI audits had previously uncovered critical vulnerabilities in the V5 proving system, including a bug that could allow forged proofs to pass. Although Aztec patched these issues in V6, he emphasized the ongoing need for rigorous testing, especially in a landscape where cyberattacks are becoming more automated and widespread. He described the current environment as one where 'the marginal cost of cyber attacks is falling to zero' and attackers can deploy 'millions of AI bots.'
Only the most security-critical parts of Aztec's protocol have been formally verified, Williamson noted. He explained that after AI found critical bugs, the team formally verified those subsystems, implying that the identified bugs were the last in those specific areas. He also mentioned that the team rarely encounters model refusals because it focuses on hunting both cryptography and ordinary software bugs, using uncensored open-source models for double-checking.
When asked about the Zcash bug reported in June, Williamson praised Zcash's risk management and explained that Aztec adopts a similar approach by performing major network upgrades as complete hard forks. This ensures that any hidden double-spend attacks would be detectable during withdrawals and deposits, preventing them from extending to the new version of Aztec. ETHBEAT (formerly L2BEAT) has rated V6 as Stage 2, its top tier, indicating a fresh start with no state carried over from V5.