Key facts
- Researchers used Anthropic's Claude Mythos AI to help build a macOS kernel exploit.
- The exploit targeted Apple's new M5 chip protections and bypassed Memory Integrity Enforcement.
- The exploit allows a local user account to gain root access on a Mac.
- The exploit is a local privilege escalation chain, likely following a malicious download or compromised installer.
- The development raises concerns about AI accelerating the discovery of software flaws by cybercriminals.
Apple devices are known for their robust security features, including protections on their latest M5 chips. However, the rapid development of AI tools is beginning to challenge these defenses. The ability of AI to assist in finding and exploiting vulnerabilities at an accelerated pace presents a significant shift in the cybersecurity landscape, potentially enabling malicious actors to discover and leverage flaws before they can be patched.