A security research team working on Apple’s M5-based Macs says Anthropic’s Mythos Preview helped them identify bugs and develop a working kernel memory corruption exploit quickly. The outlets report that the researchers bypass Apple’s memory-safety mitigation, Memory Integrity Enforcement (MIE), which is designed to make memory corruption exploits harder to execute. According to the team’s account, a key practical sequence involves running a command as a standard user and obtaining root (administrator) access. 9to5Mac and related coverage say the researchers have a detailed technical report but do not plan to publish it until Apple releases a fix for the vulnerability. The reports also say Mythos was used during the collaborative exploit development process to help recognize and work through known bug classes, and that pairing AI with human expertise was important for overcoming the challenges of bypassing a newer mitigation. TechRadar characterizes the effort as an example of how fast security problems can be investigated when AI tools are used alongside experts. Bruce Schneier’s summary similarly focuses on the team’s use of Mythos to find and exploit a kernel memory corruption vulnerability on Apple silicon.