An advanced AI security audit of the privacy-preserving cryptocurrency Zcash has concluded with no new serious vulnerabilities discovered. According to a statement by Zcash founder Zooko Wilcox, the audit, conducted by Anthropic’s Claude Mythos artificial intelligence model, confirmed the protocol’s integrity following an emergency patch implemented earlier this month.
The security review was requested by Shielded Labs, a Swiss non-profit supporting Zcash development. It followed a June 3 incident where developers temporarily suspended Orchard shielded pool transactions to patch a four-year-old forgery bug. Discovered by researcher Taylor Hornby using Anthropic’s Claude Opus 4.8 model, the vulnerability was resolved the same day with no evidence of exploitation or unauthorized value creation.
While the Mythos model successfully validated the Zcash fixes, the rapid evolution of AI auditing tools has triggered widespread industry concern. Anthropic recently disclosed that its Mythos model uncovered over 10,000 critical software vulnerabilities prior to public release, prompting the company to restrict its cybersecurity capabilities. Amid national security concerns, the US government issued export controls that forced Anthropic to suspend access to its Fable 5 and Mythos 5 models.
Cybersecurity experts warn that these powerful tools are shifting the landscape in favor of malicious actors. Immunefi CEO Mitchell Amador described the surge in AI-driven exploit identification as a “vulnerability apocalypse,” which has fueled a sharp resurgence in decentralized finance (DeFi) breaches. This structural threat is underscored by recent DefiLlama data, which shows global crypto hacks spiking to $634 million in April.
