Apple Mac M5 System Exploited With Anthropic's Claude Mythos AI, Researchers Claim

▤ Summary

A security startup, Calif, says researchers used a preview of Anthropic's Claude Mythos AI to help build a working macOS kernel exploit targeting Apple M5 systems protected by Memory Integrity Enforcement. Calif reports it discovered the attack path on April 25 and developed a working exploit by May 1, then briefed Apple in person. The company says Mythos Preview helped identify vulnerabilities and assist exploit development, but human expertise was still required to bypass MIE protections. The disclosure highlights how advanced AI tooling can accelerate vulnerability discovery and raises concerns about how mitigation technologies will hold up as models improve.

▦ Linked Companies