In just five days! Anthropic's top-secret Mythos model breaks through Apple’s five-year defense with M5 memory protection

Security research team Calif announced that they used Anthropic's unreleased Mythos Preview model to successfully build the first publicly available macOS kernel memory corruption exploit chain on a Mac device equipped with an M5 chip. MIE (Memory Integrity Enforcement) is Apple's flagship hardware security mechanism designed for M5 and A19 chips, with the company investing five years and billions of dollars into it. Its original purpose was not to be completely immune to hackers but to significantly increase exploitation costs to mitigate memory corruption vulnerabilities. Calif's team took only five days from discovering the flaw to completing the exploit.
This attack chain involves two vulnerabilities and multiple techniques, starting from a non-privileged local user, relying solely on standard system calls, ultimately gaining root access to the device. The exploit chain is a purely data-driven local kernel privilege escalation targeting macOS 26.4.1 running on bare metal hardware with the kernel MIE mechanism enabled.
The Mythos model excels at quickly generalizing from learning a specific type of attack to similar issues. It helped the team rapidly identify flaws belonging to known vulnerability classes, with human experts later overcoming new hardware defenses. This breach verification demonstrates the high efficiency of the "AI discovering vulnerabilities + expert bypass" approach, and also proves that with the assistance of top-tier large models, small security teams can challenge the technical barriers built by large companies spending heavily. $ETH
{spot}(ETHUSDT)