AI Tooling Used to Build macOS Kernel Exploit Bypassing Apple MIE
Security research firm Calif used its "Mythos" AI tooling to construct a macOS kernel memory corruption exploit that successfully bypassed Apple's Memory Integrity Enforcement during April 2026 tests, per reporting by The Wall Street Journal. The exploit chain went beyond fuzz-harness generation into genuine kernel exploitation — the first publicly reported LLM-assisted kernel-level attack against a major operating system.
Why It Matters
This moves LLM-assisted offensive security from research to demonstrated proof. Security teams should treat AI-augmented kernel exploit development as an active threat vector, not a theoretical scenario, and accelerate patch cycles for kernel-adjacent attack surfaces.