Anthropic Expands Claude Mythos Access to Critical Infrastructure Across 15 Nations

Anthropic announced on Tuesday that it is scaling up Project Glasswing, an industry-wide collaboration aimed at leveraging artificial intelligence to detect and remediate critical software flaws. The initiative is now extending to approximately 150 additional organizations spanning more than 15 countries. This expansion follows Anthropic’s confidential filing for an initial public offering just one day prior, a move that comes in the wake of a $65 billion funding round that valued the company at nearly $1 trillion.

At the core of Project Glasswing is Claude Mythos, which Anthropic describes as its most advanced model to date. The AI system has demonstrated the capacity to identify thousands of zero-day vulnerabilities within a matter of weeks. In early April, the company granted early access to Claude Mythos Preview for its initial cohort of 50 partners, which included U.S. government entities, allowing them to scan their codebases for security weaknesses.

Today’s expanded roster broadens the scope to include sectors such as power, water, healthcare, communications, and hardware—industries that Anthropic stated were underrepresented in the program's first phase. According to a blog post, many of the new participants are companies or nonprofits whose codebases are relied upon by other organizations and governments.

“What each partner has in common is that a successful attack on their codebase could be catastrophic,” Anthropic noted. “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.”

The Financial Times, citing a source familiar with the matter, reported that the participating nations include those aligned with the U.S., such as Australia, Canada, France, Germany, Italy, Switzerland, the Netherlands, Spain, Belgium, Sweden, India, Japan, New Zealand, and South Korea. The report also identified several specific entities granted access to Mythos, including U.S. identity and security management firm Okta; South Korean tech giants Samsung, SK Hynix, and SK Telecom; NATO, the Brussels-headquartered military alliance; and ENISA, the European Union’s cybersecurity agency. TechCrunch has contacted Anthropic for confirmation.

Anthropic has indicated that it anticipates other AI firms will soon release models with capabilities comparable to Claude Mythos Preview. Consequently, the company is accelerating efforts to establish robust safeguards within Project Glasswing. This competitive landscape includes rival OpenAI, which recently launched GPT-5.5-Cyber, a cybersecurity-specific model currently being tested by a large group of partners.

Source: TechCrunch Generated at: 2026-06-02 14:44:21 UTC