Calif Claims Anthropic Mythos AI Exploits Apple M5 Mac Kernel, Bypassing MIE

Security firm Calif claims it used a preview of Anthropic’s Claude Mythos AI to develop a working macOS kernel exploit on M5 hardware, bypassing Memory Integrity Enforcement (MIE) protections, with bugs found on April 25 and exploit ready by May 1. The findings, chaining two vulnerabilities for memory corruption and privilege escalation in macOS 26 and iOS, were first detailed in a WSJ report on May 14 after April tests. Apple is reviewing the disclosure, employs frontier AI for its own vulnerability testing, and has not publicly commented, amid warnings that AI accelerates bug discovery—previously flagging 271 Firefox issues—raising operational and reputational risks for AAPL.OQ though market reaction remains muted.