Federal Agencies Phase Out Anthropic AI Under Trump Executive Order

The directive issued on March 5, 2026, marks a definitive shift in the U.S. government’s approach to artificial intelligence procurement, moving away from the safety-centric 'Constitutional AI' model championed by Anthropic. By ordering federal agencies to begin a structured phase-out of Anthropic’s Claude models, the Trump administration is signaling a preference for AI architectures that prioritize raw performance and 'unfiltered' output over the rigorous safety guardrails that have defined Anthropic’s market position. This move follows months of escalating tension between the administration and tech firms that emphasize AI alignment and risk mitigation, which some officials have characterized as ideologically restrictive.

For Anthropic, the loss of the federal market is a significant blow to its enterprise growth strategy. The company had recently been in high-level negotiations with the Pentagon to integrate Claude into defense-tech workflows, ranging from logistics optimization to intelligence analysis. The abrupt termination of these prospects not only impacts immediate revenue but also creates a 'chilling effect' for other SaaS providers who have built their value propositions around ethical AI and safety compliance. While Anthropic remains heavily backed by Amazon and Google, the loss of the 'federal seal of approval' may complicate its efforts to secure similar high-security contracts in international markets and allied nations.

“

government’s approach to artificial intelligence procurement, moving away from the safety-centric 'Constitutional AI' model championed by Anthropic.

The competitive landscape is already shifting in response to the order. Reports indicate that OpenAI is positioned as a primary beneficiary, with several agencies already exploring a transition to GPT-based tools that are perceived as more aligned with the administration's current policy goals. Furthermore, Elon Musk’s xAI is expected to see increased adoption within federal circles, particularly in departments that prioritize rapid deployment and minimal output filtering. This creates a bifurcated market where AI vendors may feel pressured to choose between 'safety-first' or 'performance-first' development paths to satisfy different political and regulatory environments.