Overview
U.S. officials have signaled opposition to Anthropic’s recent proposal to widen access to its Mythos artificial intelligence model, according to people familiar with the matter. The company that produces the Claude family of models proposed granting access to roughly 70 additional companies and organizations, increasing the number with access to about 120.
Security concerns cited
White House staff have raised objections rooted in national security considerations. Anthropic itself has described Mythos as capable of identifying software vulnerabilities and executing broad cyber attacks - attributes that the company says justify a restricted rollout. That technical profile has put Washington in a consultative role during the model’s deployment, with officials watching how Anthropic plans to control distribution and use.
Compute capacity and government access
Beyond the security profile of the model, officials are worried Anthropic may lack sufficient computing power to support expansion to the proposed number of external users without undermining the government’s ability to access Mythos when needed. That capacity concern is central to the administration’s resistance to a larger set of licensees.
No public release planned
Anthropic has not indicated any intention to open Mythos to the general public. Instead, the company’s rollout remains restricted, in part because of the potential national security implications it has identified.
Funding and compute plans
Separately, reports indicate Anthropic is in discussions with investors about raising capital at a reported $900 billion valuation. Those fundraising efforts are said to include a component aimed at securing additional computing resources to support Mythos - an objective that aligns with the company’s stated need for more capacity to serve users securely.
Market tickers noted in reporting
The article referenced the following market tickers in ancillary market data: GOOGL, AMZN, NVDA.