Odaily, citing Cointelegraph, reported that Zcash founder Zooko Wilcox said in a post on X that a security audit performed by Anthropic’s Claude Mythos artificial intelligence model did not discover “more severe vulnerabilities” in the Zcash protocol. The audit was requested by Shielded Labs, a Swiss nonprofit organization that supports the development of Zcash.
Shielded Labs requested the Claude Mythos review
Wilcox’s statement focused on the outcome of the Claude Mythos review: the model’s audit did not identify a more serious flaw in the Zcash protocol. The disclosure follows a recent period in which Zcash developers and related organizations have publicly addressed protocol security issues, including the handling of a flaw in the Orchard shielded pool.
On June 3, Zcash developers temporarily paused Orchard transactions after identifying a shielded pool vulnerability. Functionality was restored the same day through an emergency upgrade. The issue stemmed from a forgery vulnerability in the Orchard shielded pool that had existed for four years.
The Orchard issue was found with help from Claude Opus 4.8
The Orchard shielded pool forgery vulnerability was discovered by security researcher Taylor Hornby with assistance from Anthropic’s Claude Opus 4.8 model. The Zcash Foundation said there was no evidence that the vulnerability had been exploited, no unauthorized value creation had been detected, and user privacy was not affected.
Anthropic released the first public version of its Claude Mythos model, Fable 5, on Tuesday. On Friday, the company said it had suspended access to the Fable 5 and Mythos 5 AI models after the U.S. government issued export control instructions citing national security concerns.