A four-year hidden vulnerability was uncovered by Claude; AI auditing is really not just a gimmick.

BitBean News reports that after the Orchard vulnerability was fixed, the Zcash founder said that AI-assisted security audits did not find any new critical vulnerabilities. No additional flaws were revealed by Shielded Labs’ commissioned review by Anthropic’s Claude Mythos. After it was discovered on June 3, developers paused related transactions and restored operations through an emergency upgrade. The vulnerability existed for about four years, and it was discovered by Taylor Hornby with the assistance of Claude Opus 4.8. The foundation said there is no evidence that it was exploited or that user data was leaked. This case highlights both the role and potential risks of AI in cryptocurrency security.