Anthropic Expands ‘Project Glasswing’ to Secure Critical Infrastructure Anthropic is accelerating its push to use advanced AI to defend critical infrastructure, even as the scale of the effort raises fresh questions about concentrating so much cyber power in a single model.

Early trials with Claude Mythos

In early April, Anthropic quietly launched “Project Glasswing,” giving roughly 50 initial partners, including the U.S. government, access to a preview of its most powerful security-focused AI model, Claude Mythos. Those partners used the model to scan their codebases and, according to Anthropic, uncovered “more than 10,000 high- or critical-severity security flaws” in a matter of weeks.

June expansion across 15+ countries

On June 2, Anthropic announced it is expanding Project Glasswing to about 150 additional organizations in more than 15 countries, many of them in sectors like power, water, healthcare, communications, and hardware that were “not well represented” in the initial cohort. Many of the new partners are vendors or nonprofits maintaining software that governments and companies around the world depend on.

Anthropic stresses the stakes: “What each partner has in common is that a successful attack on their codebase could be catastrophic,” with a major breach potentially affecting “more than 100 million people” and carrying “important ramifications for both global and national security.”

Industry and geopolitical context

From the AI developer’s perspective, Project Glasswing is a “collaborative effort to secure the world’s most important software” and to help cybersecurity adapt as “cheap, fast AI models with powerful cyber capabilities are around the corner.” Human analysts note the expansion coincides with Anthropic’s confidential IPO filing after a massive funding round and valuation near $1 trillion, underscoring how central security applications have become to its business strategy.

Tech reporting also highlights that access to Mythos is being extended primarily to organizations in U.S.-aligned countries and major firms like Okta and Samsung, as well as NATO and the EU’s cybersecurity agency, reflecting both commercial ambitions and geopolitical alignment in how this AI cyber capability is being deployed.

Continue reading https://foxvector.com/stories/019e8953-de3e-2691-7322-39dce3713a31