Trump signs executive order barring 'woke AI' from federal procurement

President Trump signed “Preventing Woke AI in the Federal Government,” one of three executive orders issued alongside the administration’s AI Action Plan. The order directs federal agencies to procure only large language models that comply with two “Unbiased AI Principles”: that they be “truth-seeking,” prioritising historical accuracy and objectivity over preferred outcomes, and “ideologically neutral,” functioning as tools that do not encode partisan or ideological judgments unless a user deliberately elicits them.

The order frames the problem as models built with “diversity, equity, and inclusion” concepts embedded at the expense of accuracy, citing as examples chatbots that depicted historical figures with altered races or declined to acknowledge the achievements of particular groups. It directs the Office of Management and Budget to issue implementation guidance within 120 days, after which agencies have 90 days to build compliance terms into new contracts and, where practical, existing ones. Noncompliant vendors face contract termination, with decommissioning costs charged to them; national-security systems can be exempted, and vendors may satisfy the neutrality requirement through disclosure of a model’s judgments rather than technical changes.

Civil-liberties and AI-policy critics argued the order used the federal government’s purchasing power to pressure vendors into altering commercial products well beyond government use, and that “ideological neutrality” was not a testable technical property, leaving compliance largely to negotiation between agencies and vendors. Supporters framed it as removing a specific form of bias they said had already been demonstrated in deployed chatbots. Its practical effect depended on the OMB guidance that followed, since the order itself set principles rather than a compliance test.