Mythos AI Helps Researchers Crack Apple M5 Kernel Security in 5 Days
Original: Elite researchers teamed up with Anthropic's Mythos AI to smash Apple's multi-billion dollar M5 security and build a kernel exploit in just 5 days. View original →
What Happened
Researchers at Calif, working alongside Anthropic's Mythos Preview AI, developed the first publicly disclosed macOS kernel memory corruption exploit that bypasses Apple M5's Memory Integrity Enforcement (MIE). The timeline: initial bug identification on April 25th, working exploit by May 1st — five days.
Why MIE Matters
MIE is Apple's flagship hardware-assisted memory safety system, built on ARM's Memory Tagging Extension, developed over five years at estimated cost in the billions. Apple's own research shows MIE disrupts every known public exploit chain against modern iOS, including the recently leaked Coruna and Darksword exploit kits. That is what was bypassed.
AI's Role
Mythos Preview rapidly identified bugs within known vulnerability classes. Human experts provided the specialized knowledge required to circumvent MIE's advanced protections. The team explicitly credits this human-AI collaboration with enabling a small group to achieve in one week what previously required large, well-resourced organizations.
The Bigger Picture
This isn't just a security disclosure — it is a demonstration of AI-accelerated vulnerability research capability. The same tools that enabled this responsible disclosure could be applied offensively. The team delivered a 55-page technical report directly to Apple Park. The defense side of the equation now has to contend with dramatically shortened attacker timelines.
Related Articles
Why it matters: the same model Anthropic framed as too dangerous for public release was reportedly exposed twice in quick succession. The Verge says Mythos was first revealed through an unsecured data trove, then reached by unauthorized users from day one through guessed infrastructure and contractor access.
The Future of Life Institute’s Summer 2026 AI Safety Index grades nine frontier AI companies across 37 indicators, and no firm rises above C+. The sharper point is not who leads, but how weak the ceiling remains as model capabilities and defense use expand.
Anthropic said Claude Opus 4.6 found 22 Firefox vulnerabilities during a two-week collaboration with Mozilla. Mozilla classified 14 as high severity and shipped fixes in Firefox 148.0.