Anthropics Mythos Begins Detecting Security Flaws in Apple Software

Anthropic’s Claude Mythos is an AI recognized for its expertise in detecting security vulnerabilities. The company chose to keep it accessible only to a specific group of security researchers and large organizations rather than the general public. Recently, its capabilities are being showcased through its limited release to partners like Apple, where its effectiveness in security was demonstrated by identifying a flaw in Apple’s macOS, according to The Wall Street Journal. Researchers from Calif found the vulnerability, marking it as the first public exploit of its kind on Apple M5, which could allow unauthorized access to the device. The discovery involved two vulnerabilities and numerous techniques, with Mythos assisting significantly. Although it’s not verified if the flaw has been addressed yet, MacRumors noted Apple’s recent macOS Tahoe 26.5 update, mentioning a fix submitted by Calif in collaboration with Claude and Anthropic Research. Calif suggested forthcoming technical details post-resolution by Apple, emphasizing Anthropic’s role in the findings. An Apple representative reaffirmed their commitment to security in response to the report.