FAIR Act
Introduced September 11, 2025 · Last action September 11, 2025
Plain English Summary
This bill requires federal agencies to purchase only large language models (AI chatbots) that are developed according to specific principles: they must be truthful, historically accurate, scientifically objective, and "neutral" without what the bill calls ideological bias toward diversity, equity, and inclusion initiatives. Developers cannot embed partisan or ideological judgments into the AI unless users can easily access and modify them.
Who benefits
AI developers and companies that build large language models without diversity, equity, and inclusion safeguards or that market themselves as neutral and objective alternatives to existing AI systems; conservative-aligned or right-leaning AI companies positioning themselves against DEI-focused competitors; federal agencies seeking AI tools aligned with this bill's ideological framework; Republican members of Congress and conservative constituencies opposing DEI initiatives in government technology procurement.
Who pays / loses
AI companies that currently incorporate diversity and inclusion considerations into model training and deployment (including major cloud providers like OpenAI, Google, Microsoft, and Anthropic); federal agencies that have adopted widely-used commercial LLMs with DEI-aligned safeguards and must now find or commission alternatives; government contractors and vendors dependent on selling mainstream commercial AI; technology companies seeking federal contracts for AI services.
Funding & Lobbying Interests
Conservative technology advocates and companies marketing AI as an alternative to DEI-focused models have financial incentive in this bill's passage. Rep. Hageman represents Wyoming, a state with strong Republican representation and energy industry ties; conservative donors opposing ESG (environmental, social, governance) and DEI frameworks in business would benefit from legislation restricting federal procurement of mainstream AI. The bill lacks explicit funding mechanisms or appropriations, suggesting it operates as a procurement restriction rather than a funded program, potentially benefiting smaller AI startups or conservative-oriented tech firms over large incumbent cloud providers (Microsoft, Google, Amazon, OpenAI) that dominate federal AI procurement.
Political Impact
Affected Groups
Federal government agencies and their IT procurement offices (all cabinet departments, military, intelligence agencies, and government corporations); AI and cloud computing companies competing for federal contracts; federal employees who depend on AI tools for job functions; taxpayers funding federal AI procurement; technology workers at companies whose products are excluded; users of federal services relying on AI-assisted decision-making.
Political Subtext
Proponents argue this bill protects government AI from ideological bias and ensures taxpayer money funds objective, trustworthy AI systems. They contend DEI-aligned AI models manipulate responses to advance progressive political goals and that federal agencies should only buy neutral tools. Critics argue the bill codifies a narrow conservative definition of neutrality and objectivity, mischaracterizes how modern AI safety training works, and creates a political litmus test for federal procurement. They note that all AI systems reflect choices by developers and that framing DEI safeguards as inherently biased while requiring models without them is itself an ideological choice. Non-partisan evidence shows no consensus definition of "bias-free" AI exists; researchers across the political spectrum debate whether removing DEI-focused training produces more or less biased outputs. The bill does not specify how agencies will verify compliance or define compliance thresholds.
Real-World Stakes
If this passes: Federal agencies would be prohibited from using widely-adopted commercial AI systems like ChatGPT, Claude, Gemini, or similar models that incorporate diversity and safety safeguards. Agencies would need to identify, commission, or develop alternative AI systems or go without LLM technology. This could delay or complicate federal AI adoption compared to private sector capabilities. It creates a precedent for federal procurement to restrict technology based on ideological requirements rather than technical merit or cost-effectiveness. Analogous state-level DEI restrictions (e.g., Florida and Texas laws restricting DEI in state contracting, passed 2023-2024) have led to vendor exits from those states, reduced competition, and higher procurement costs when available alternatives are limited. The bill does not mandate an alternative procurement source, risking federal reliance on smaller, less-tested AI systems or the emergence of a carve-out vendor market serving only conservative-acceptable criteria.
What's Inside This Bill
Non-obvious provisions not apparent from the headline summary.
Undefined 'Ideological Bias' Enforcement Standard
The bill requires AI systems to be free of 'ideological bias toward diversity, equity, and inclusion initiatives' but does not define what constitutes such bias or establish objective metrics for compliance. This shifts enforcement discretion to federal agencies with no clear technical or measurable standards.
AffectsAI developers will face unpredictable compliance requirements and potential contract rejection based on subjective agency interpretation. Small AI companies lacking lobbying resources may be disproportionately affected by inconsistent enforcement across agencies.
Broad Federal Procurement Ban Without Transition Period
The bill appears to require immediate cessation of purchases of non-compliant LLMs across all federal agencies without specifying implementation timelines, phase-in periods, or exceptions for existing contracts and systems.
AffectsFederal agencies currently using mainstream AI tools (like ChatGPT, Claude, Gemini) for government operations may face service disruptions. Agencies without procurement flexibility could experience operational delays in adopting replacement systems.
'User Modification' Requirement Without Technical Specification
The bill requires developers to make ideological judgments 'easily accessible' and modifiable by users, but does not define what 'easily' means technically, who qualifies as a 'user,' or what 'modification' entails, potentially creating compliance ambiguity.
AffectsAI vendors must guess at implementation standards, and government agencies must determine who within their organizations can modify AI behavior, potentially creating security, liability, and audit concerns for sensitive government use cases.
Exclusion of Specific LLM Characteristics Without Delineation
The bill prohibits 'partisan or ideological judgments' but does not clarify whether this applies to training data selection, model architecture choices, or output filtering—creating uncertainty about what development practices violate the standard.
AffectsAI developers cannot determine which standard industry practices (dataset curation, fine-tuning, constitutional AI methods) remain permissible, potentially chilling innovation in responsible AI development practices.
Sponsor
Co-sponsors (1)
RRep. Moore, Barry [R-AL-1]
Vote Record
No recorded votes.
Campaign Finance — Primary Sponsor
Top contributing industries
Other$396,871.39
Finance$8,290.28
Energy$8,016.65
Law$4,525.51
Transportation$4,331.84
501(c)(4) disclosure: Contributions from 501(c)(4) "dark money" organizations are not required to be publicly disclosed and are not reflected in the figures above. Data sourced from FEC public disclosure filings.
Community Discussion
Share this bill
Sign in to join the discussion.
No comments yet. Be the first.