Anthropic Expands Access to Claude Mythos After AI Giant Files for IPO
Anthropic is expanding controlled access to its Claude Mythos model through Project Glasswing, adding about 150 organizations from critical infrastructure, security, open-source, and government-adjacent sectors. The goal is to let trusted partners find bugs, vulnerabilities, and exploit risks before a wider release. Anthropic says current partners have already uncovered more than 10,000 high- or critical-severity vulnerabilities and are using the model for patching, security reviews, and pre-release testing. The company argues defenders need advanced AI tools before attackers get similar capabilities, especially as other frontier models are expected within 6–12 months. The move follows reports that Mythos can autonomously carry out complex cyberattack simulations, reinforcing concerns that stronger AI could lower the barrier to cyberattacks.
