Anthropic Expands AI Vulnerability Tool to 150 Organizations

Anthropic is expanding Project Glasswing, its AI-powered initiative to detect critical software vulnerabilities, to 150 new organizations across 15+ countries. The program uses Claude Mythos, the company's most powerful model, capable of identifying thousands of zero-day vulnerabilities. New partners span power, water, healthcare, communications, and hardware sectors. Organizations now included are Okta, Samsung, SK Hynix, SK Telecom, NATO, and the EU cybersecurity agency ENISA. Anthropic estimates a successful attack on most partners' codebases could affect over 100 million people. The expansion comes as rival OpenAI launches its own cybersecurity model, GPT-5.5-Cyber.