This executive order represents a significant shift in federal AI procurement policy, positioning itself as a corrective measure against what it characterizes as ideological bias in artificial intelligence systems. The order frames diversity, equity, and inclusion (DEI) principles as fundamentally incompatible with truthful AI outputs, arguing that such ideologies "displace the commitment to truth in favor of preferred outcomes" and pose "an existential threat to reliable AI." Building upon Executive Order 13960 from 2020, this directive establishes the administration's intent to prevent federal agencies from procuring AI systems that allegedly prioritize social agendas over factual accuracy. The order cites specific examples of AI models that it claims altered historical representations or refused certain requests based on DEI considerations, framing these as evidence of systematic distortion in AI outputs. Notably, the order's scope may extend beyond procurement to agency-developed AI systems and non-LLM technologies, pending OMB guidance that will specify factors for such expansion.
The order establishes two mandatory "Unbiased AI Principles" that must govern all federal procurement of large language models (LLMs). The first principle, "Truth-seeking," requires LLMs to prioritize "historical accuracy, scientific inquiry, and objectivity" while acknowledging uncertainty when information is incomplete. The second principle, "Ideological Neutrality," mandates that LLMs function as "neutral, nonpartisan tools" that do not manipulate responses to favor what the order terms "ideological dogmas such as DEI." However, these requirements present significant operational and legal risks, including challenges to defining and enforcing "ideological neutrality" in contested domains, potential conflicts with existing civil rights mandates, and feasibility concerns around implementing "truth-seeking" standards for complex or ambiguous topics. The contentious nature of these principles may trigger litigation and compliance challenges that could complicate implementation across diverse federal agencies.
Implementation responsibility falls primarily to the Office of Management and Budget, which must issue comprehensive guidance within 120 days in consultation with federal procurement and technology officials. Agency heads are then required to incorporate compliance terms into all future LLM contracts, including provisions that charge vendors for decommissioning costs if contracts are terminated for non-compliance. While the order allows for technical limitations and vendor flexibility, the stringent requirements and potential penalties may significantly impact the federal government's relationships with AI vendors, potentially deterring leading providers, increasing procurement costs and timelines, or limiting access to cutting-edge AI technologies. Each agency must adopt internal procedures within 90 days of OMB guidance, fundamentally reshaping how the federal government evaluates and contracts for AI services while potentially constraining its competitive position in the AI marketplace.