Anthropic's 'Claude Mythos' AI Model Uncovers Security Flaws in Every Major OS and Browser

Anthropic has unveiled a new AI model, Claude Mythos Preview, that reportedly identified security vulnerabilities 'in every major operating system and web browser.' This discovery is part of Project Glasswing, a high-stakes cybersecurity initiative launched in partnership with tech titans including Nvidia, Google, Amazon Web Services, Apple, and Microsoft. The model is designed to autonomously flag critical system vulnerabilities for large corporations and potentially government entities, aiming to operate with minimal human oversight.

The model is not slated for public release due to significant security concerns, with access currently restricted to the elite consortium of launch partners. According to Newton Cheng, Anthropic's frontier red team cyber lead, Claude Mythos represents a shift toward using general-purpose AI for offensive security research at scale. The initiative signals a new phase where foundational AI models are directly deployed to probe and harden the digital infrastructure of the world's most powerful tech firms.

The closed nature of the project and the caliber of its partners underscore the strategic and sensitive value of the findings. This collaboration places immense pressure on the involved operating system and browser vendors to rapidly address the flagged flaws. The model's capability suggests a future where AI-driven discovery could outpace traditional human-led security teams, reshaping vulnerability management and defense postures across the entire technology sector.