The U.S. Office of Management and Budget (OMB) recently released detailed guidance directing federal agencies to ensure the AI systems they procure comply with the Trump administration’s mandate against so-called “woke AI.” 

The guidance lays out how federal agencies must update procurement policies, structure new contracts, and amend existing ones to verify compliance.
Vendors must provide “sufficient information” to demonstrate their models align with the administration’s principles, though agencies are instructed not to request sensitive technical data such as model weights. Acceptable documentation may include training summaries, evaluation scores, developer resources, and mechanisms for user feedback.
The seven-page memo, issued by OMB Director Russell Vought, follows President Donald Trump’s July executive order titled “Preventing Woke Al in the Federal Government” requiring models used by the federal government to be “truth-seeking” and “ideologically neutral.”

“Agencies must also consider the relevant factors identified in determining whether to apply these requirements to LLMs developed by the agency, and to AI models other than LLMs that are procured by the agency,” OMB said.

The memo also acknowledges the challenges of transparency when AI tools are purchased through resellers. While the administration argues the rules will reduce bias, researchers and industry experts have questioned whether genuine neutrality in AI is technically achievable or whether the policy risks constraining U.S. competitiveness.