On July 23, 2025, President Trump signed a new executive order—Preventing Woke AI in the Federal Government—that prohibits federal agencies from procuring or promoting large language models (LLMs) exhibiting ideological bias linked to diversity, equity, and inclusion (DEI) principles. The order mandates that LLMs used by the federal government must follow “Unbiased AI Principles,” requiring both truth-seeking and ideological neutrality
What Counts as “Woke AI”?
Under the order, any AI model promoting ideologies like critical race theory, unconscious bias, or social justice agendas will be disqualified from federal contracts. The White House cites documented cases where AI-generated images altered historical figures’ race or gender to prioritize DEI over historical accuracy
How It Will Be Enforced
- The Office of Management and Budget (OMB) must issue procurement guidance within 120 days
- Federal agencies must then include compliance requirements in new contracts for LLMs—and retrofit existing ones
- Vendors may be held financially responsible for noncompliance
Part of a Broader AI Policy Shift
This executive order comes as part of a suite of three AI-related directives signed the same day:
- Woke AI Ban – Blocks ideologically loaded LLMs from federal use
- Datacenter Deregulation – Speeds up permitting and removes DEI and environmental restrictions for new AI infrastructure
- AI Export Initiative – Promotes U.S. AI exports through grants and partnerships
Why It Matters
- The U.S. government is one of the largest buyers of AI technology—this policy essentially sets a new bar for acceptable bias in federally funded AI Axios
- Critics caution that “truthfulness” and “neutrality” are subjective—some worry this could politicize AI development and grant the government overreach in tech adjudication
✅ Bottom Line
Trump’s executive order banning “woke AI” models marks a significant shift in Federal AI procurement policy. It demands new neutrality standards for government-used models, streamlines infrastructure approvals, and promotes U.S. AI exports. But it also sparks debate about who gets to define bias and whether this move could politicize technology.
