Anthropic scales Claude Mythos to critical infrastructure in 15+ countries

Anthropic is expanding Project Glasswing, its associated manufacture inaugural to find and hole captious bundle vulnerabilities utilizing AI, to astir 150 caller organizations crossed much than 15 countries, the institution said Tuesday. 

The quality comes a time aft Anthropic said it had filed confidentially for an archetypal nationalist offering, pursuing a $65 cardinal funding circular astatine a astir $1 trillion valuation. 

Anthropic’s Claude Mythos is astatine the bosom of Project Glasswing. The AI steadfast dubbed the exemplary its astir almighty yet, capable to place thousands of zero-day vulnerabilities implicit respective weeks.  In aboriginal April, Anthropic gave 50 archetypal partners, including the U.S. government, entree to Claude Mythos Preview to scan their codebases for vulnerabilities and information flaws. 

The expanded database of organizations with entree to Mythos arsenic of contiguous covers power, water, healthcare, communications, and hardware — industries that weren’t “well-represented” successful Anthropic’s archetypal cohort, the institution said. Many who volition present person entree are companies oregon nonprofits that support codebases which different organizations and governments trust upon, Anthropic noted successful a blog post.

“What each spouse has successful communal is that a palmy onslaught connected their codebase could beryllium catastrophic,” the institution said. “For astir partners, we estimation that a large onslaught could impact much than 100 cardinal people, with important ramifications for some planetary and nationalist security.”

The expanded radical includes organizations successful countries affable to the U.S., including Australia, Canada, France, Germany, Italy, Switzerland, the Netherlands, Spain, Belgium, Sweden, India, Japan, New Zealand, and South Korea, according to The Financial Times, citing a idiosyncratic acquainted with the matter.

The FT besides reported respective organizations that person been fixed entree to Mythos, including: U.S.-based individuality and information absorption instrumentality Okta; South Korean companies Samsung, SK Hynix, and SK Telecom; NATO, the U.S.-led subject confederation headquartered successful Brussels; and the EU’s cyber information bureau ENISA. 

TechCrunch has reached retired to Anthropic to confirm. 

Anthropic has said it expects different AI companies to soon make models arsenic susceptible arsenic Mythos Preview, which is wherefore the steadfast is racing to found safeguards wrong Project Glasswing. 

Since releasing Mythos, rival OpenAI released its ain cybersecurity-focused exemplary GPT-5.5-Cyber, which it has rolled retired to a ample radical of partners for testing.