Anthropic has expanded its Project Glasswing initiative, which uses AI to identify and resolve critical software vulnerabilities, to about 150 new organizations across more than 15 countries. The company said the expansion includes key sectors such as power, water, healthcare, communications, and hardware. According to the company, the new partners are organizations that maintain codebases essential to other entities and governments. "What each partner has in common is that a successful attack on their codebase could be catastrophic," Anthropic stated.
"For most partners, we estimate that a major attack could affect more than 100 million people, with important ramifications for both global and national security." Anthropic’s Claude Mythos model is central to the initiative, with the company calling it its most powerful yet, capable of identifying thousands of zero-day vulnerabilities over several weeks. The expanded access to the model includes organizations in countries friendly to the U.S., such as Australia, Canada, France, Germany, and South Korea. The list also includes companies like Okta, Samsung, and SK Hynix, as well as NATO and the EU’s cybersecurity agency ENISA. TechCrunch has reached out to Anthropic for confirmation. Anthropic noted that it expects other AI companies to soon develop models as capable as Mythos Preview, which is why the firm is racing to establish safeguards within Project Glasswing.
Since releasing Mythos, rival OpenAI has rolled out its own cybersecurity-focused model, GPT-5.5-Cyber, to a large group of partners for testing. Source: techcrunch