Breaking through Apple’s five-year-developed security measure in just five days: The power of vulnerability attacks demonstrated by “Mythos” — in the security world these days

Calif said it used Anthropic’s Mythos Preview AI to build a Mac privilege-escalation exploit in about five days.

The proof of concept targeted Apple’s new Memory Integrity Enforcement, or MIE, on M5 Macs and macOS.

The company says the result shows how AI can rapidly accelerate offensive security research against advanced defenses.

The finding highlights the growing race between Apple’s protection technologies and AI-assisted attack development.