dailyai.report

The Claude Mythos model can autonomously assemble complex exploits against critical software. To mitigate this risk, Anthropic limited access to a select group of cybersecurity firms via Project Glasswing. These firms will use the model to patch vulnerabilities before malicious actors exploit them. This restriction highlights the growing tension between capability and safety.