Anthropic is reportedly on the verge of launching a preview version of Claude Mythos, its most advanced AI model to date. Development for this model commenced on April 7th, and it is specifically tailored for high-level computer security tasks. In comparison to its current flagship model, Opus 4, the Claude Mythos preview demonstrates notable enhancements in code comprehension and autonomous operation. However, given the model’s capacity to autonomously acquire professional-level cyberattack skills, its deployment could potentially pose a significant threat to digital infrastructure. Consequently, Anthropic is proceeding cautiously and has refrained from a full-scale release. Earlier, users briefly noticed an option to activate this model in the public versions of Claude, Claude Security, and Code, but it was quickly removed. As of now, Anthropic has not officially disclosed a definitive launch date for the Claude Mythos preview. Furthermore, Anthropic has confirmed the progression of a new initiative named "Glass," which seeks to partner with other entities to safeguard critical global software systems and thwart AI-driven vulnerability exploitation. The "Glass" project will leverage the upcoming Claude Mythos preview and has already initiated collaborations with several organizations. It has also demonstrated a source vulnerability panel designed to summarize vulnerabilities of varying severity levels identified by the preview version.