Federal Agencies Secretly Test Anthropic's AI Despite Trump Administration Ban

Despite a formal directive from President Donald Trump ordering federal agencies to stop using Anthropic's artificial intelligence technology, multiple U.S. government bodies have been quietly evaluating the company's latest AI model, according to a Politico report. The move signals a growing tension between executive policy and the urgent demand for cutting-edge cybersecurity tools.

Staff from several agencies, including the Commerce Department's Center for AI Standards and Innovation, have reportedly been testing Anthropic's newest model, referred to as "Claude Mythos." The system has captured significant attention within government circles due to its advanced capability to identify critical software vulnerabilities — flaws that often go undetected by even the most experienced human cybersecurity professionals.

The Trump administration's ban on Anthropic's technology stems from disputes over the company's position on military and surveillance applications. However, federal officials appear willing to work around that restriction, quietly assessing and potentially deploying the model based on its perceived value in strengthening national cyber defenses.

Interest has extended beyond the executive branch. Members of Congress and their staff have also requested briefings on the AI system, reflecting mounting concern that foreign adversaries — particularly well-resourced state actors — could develop or already be leveraging similar technologies to exploit American infrastructure.

The White House has acknowledged its continued engagement with artificial intelligence companies to assess and mitigate security risks, while emphasizing the need to balance those efforts against broader national security priorities.

This situation highlights a larger debate unfolding across Washington: as AI capabilities rapidly advance, federal agencies are increasingly struggling to reconcile restrictive policy frameworks with the practical need to stay ahead of emerging cyber threats. The quiet testing of banned technology suggests that for some officials, national security may ultimately outweigh political directives.