Anthropic said on June 2 it would expand Project Glasswing to 150 additional partners in more than 15 countries, giving vetted organizations access to Mythos Preview to help identify software vulnerabilities.

The new group includes organizations in power, water, healthcare, communications, and hardware. Anthropic said access remained limited to organizations that build or maintain software critical to the internet and other essential infrastructure.

Anthropic described Glasswing as an effort to secure critical software before similar AI tools are more widely available. The company also said Mythos-class models can shrink time and cost for finding and exploiting flaws.

Anthropic began the pilot with about 50 partners in April following concerns that the model could help attackers discover software weaknesses faster. The risk drew White House attention and led to meetings with the US government.

The company said partners had found more than 10,000 high- or critical-severity flaws since April.

Anthropic said on Monday it would begin offering Mythos access in the European Union.

🔗 Source: Anthropic

Recent Anthropic developments