Anthropic's Project Glasswing - restricting Claude Mythos to security researchers - sounds necessary to me

Anthropic has restricted access to its latest AI model, Claude Mythos, through Project Glasswing, allowing a select group of partners to test its advanced cybersecurity features. The model has already identified numerous vulnerabilities in major operating systems and browsers, prompting concerns about its potential misuse. Security experts note a significant improvement in AI-generated vulnerability reports. The initiative aims to strengthen cybersecurity before the technology's capabilities are broadly released.