Calif Finds 2 macOS Bugs Enabling Privilege Escalation With Mythos AI
Updated
Updated · 9to5Mac · May 14
Calif Finds 2 macOS Bugs Enabling Privilege Escalation With Mythos AI
11 articles · Updated · 9to5Mac · May 14
Calif delivered Apple a 55-page report detailing a privilege-escalation exploit that chains two macOS bugs with several techniques to corrupt memory and reach protected parts of a Mac.
April testing of an early Anthropic Mythos model helped researchers uncover the attack path, though Calif said the exploit also required significant human cybersecurity expertise.
Apple said it is reviewing the findings and reiterated that it takes potential vulnerabilities seriously; Calif expects the underlying bugs to be fixed quickly before releasing technical details.
If paired with other attacks, the exploit could let a hacker seize control of a computer, underscoring how advanced AI tools may accelerate discovery of high-impact software flaws.
An AI helped secure macOS, but what stops similar AI from being used by hackers for devastating attacks?
As AI finds flaws faster than they can be fixed, are we entering an age of permanent vulnerability?
AI-Powered Discovery of 2 Major macOS Security Flaws: Mythos, Project Glasswing, and Apple’s Rapid Patch
Overview
In May 2026, Calif used Anthropic’s advanced AI, the Mythos Preview model, to uncover two critical privilege escalation vulnerabilities in Apple’s macOS. This discovery shows how artificial intelligence is becoming essential for finding complex security flaws that might otherwise go unnoticed. By using AI-powered bug hunting tools, companies can strengthen their products and help prevent major security incidents before they affect users. The process combined Mythos’s ability to quickly scan large amounts of code and spot suspicious patterns with human experts who carefully reviewed and turned these findings into actionable security reports.