Project Glasswing: Securing critical software for the AI era (anthropic.com) AI

Anthropic and a consortium of major tech, security, and infrastructure companies are launching Project Glasswing to use the company’s frontier model, Claude Mythos Preview, for defensive cybersecurity. The initiative aims to help partners scan critical software for vulnerabilities and speed up patching, while Anthropic shares learnings with the broader industry and supports open-source security efforts. The announcement is driven by concerns that AI models’ coding and vulnerability-exploitation capabilities may soon scale beyond human defenders if not harnessed for protection.