AI for Business

Anthropic's Restricted AI Model Gains White House Backing for Cybersecurity

The White House is preparing to provide federal agencies with access to a powerful and controversial artificial intelligence model, marking a swift reversal of its previous stance. According to a...

Share:

The White House is preparing to provide federal agencies with access to a powerful and controversial artificial intelligence model, marking a swift reversal of its previous stance. According to a memo from federal CIO Gregory Barbaccia, officials are now establishing safeguards for Anthropic's Mythos model, with deployment to departments including Defense and Homeland Security possible in the near term. This shift comes just weeks after the prior administration labeled Anthropic a national security supply chain risk and ordered its technology purged from government systems.

The change follows the model's controlled leak and subsequent containment. Anthropic confirmed Mythos was too potent for public release, citing significant cybersecurity risks, and instead launched a less advanced public version. Select corporations, including several major banks and tech firms, received private previews, with demonstrations reportedly prompting urgent meetings at institutions like the Bank of England.

Anthropic's founders briefed U.S. officials early on the model's dual-use potential for both offensive and defensive cyber operations. This opened a path to collaboration on safety measures. The turnabout appears driven by practical necessity: internal testing revealed Mythos can identify critical software vulnerabilities—some hidden for years—in a fraction of the time. A White House-led initiative, Project Glasswing, now involves roughly 50 organizations using the model to automate the discovery and patching of digital flaws.

While civilian agencies are first in line, tensions with the Pentagon persist over Anthropic's refusal to modify its technology for certain military applications, a stance that triggered the original ban. Legal disputes continue. Nevertheless, the administration's priority is clear: deploy a guarded version of this advanced AI to fortify national cyber defenses, betting that controlled access outweighs the risk of leaving the tool entirely in the hands of adversaries or private entities.

Source: Webpronews

Ready to Modernize Your Business?

Get your AI automation roadmap in minutes, not months.

Analyze Your Workflows →