Zcash Says Anthropic AI Audit Found No Additional Vulnerabilities

Zcash founder Zooko Wilcox on June 12 announced the results of a security audit conducted with an Anthropic AI model, saying the review found no other major vulnerabilities in the protocol.

In a post on X, Wilcox wrote that the audit was carried out at the request of digital-asset research group Shielded Labs using Anthropic's AI model Mithril. The review did not uncover any other serious flaws in the Zcash protocol, he wrote.

He added that teams including Shielded Labs are continuing work to strengthen security and will provide a follow-up update soon.

The audit was part of efforts to address a vulnerability recently identified on the Zcash network. On June 5, Wilcox disclosed a counterfeiting vulnerability in Zcash's Orchard pool that could have allowed unlimited creation of fake ZEC.

That flaw was also identified using Anthropic's AI technology. Security researcher Taylor Hornby found it during a targeted audit on May 29 using Anthropic's Opus 4.8 model and reported it to Zcash Open Development Lab, or ZODL. After coordinating with participants across the Zcash ecosystem, ZODL completed a fix for the defect on June 2.