White House approves federal use of Anthropic’s Claude Mythos AI despite cybersecurity concerns and ongoing legal clashes
The White House Office of Management and Budget is implementing security protocols for federal agencies to start using Anthropic’s Claude Mythos Preview AI model.Federal CIO Gregory Barbaccia communicated this plan to Cabinet departments, as detailed in a memo obtained by Bloomberg.
AI model approved
Government officials are concerned about the model’s potential to raise cybersecurity risks amid ongoing legal disputes involving the Trump administration and Anthropic regarding the AI technology’s use.
Anthropic released Claude Mythos Preview on April 7, initially limiting access to 40 partner organisations through its Project Glasswing initiative.
During testing, engineers found that the model identified software vulnerabilities overnight and provided a comprehensive method for exploiting them.
The company reported that Mythos detected thousands of previously unknown high-severity vulnerabilities, with less than 1% of these patched.
Anthropic has stated it will not make the model publicly available.
Co-founder Jack Clark confirmed that Anthropic briefed the Trump administration on Mythos despite existing tensions.
Government oversight
Clark emphasised the importance of government awareness regarding advancements in AI technology.
The rapid evolution of AI raises critical questions about safety and regulation in various sectors.
Continued discussions between the government and AI companies will likely shape future protocols.
Comments
No comments yet. Start the conversation.