Preventing Woke AI in the Federal Government

Executive Order Number

14319

President

Donald Trump

Signed

July 23, 2025

Published

July 28, 2025

Source

Federal Register

Original Document

https://www.govinfo.gov/content/pkg/FR-2025-07-28/pdf/2025-14217.pdf

AI-Generated Summary

Below is a detailed summary and analysis of the provided executive order regarding the procurement of large language models (LLMs) by federal agencies, formatted in Markdown for clarity and readability.

---

Summary of Executive Order on Unbiased Artificial Intelligence in Federal Procurement

Purpose

• The executive order aims to ensure that artificial intelligence (AI), specifically large language models (LLMs), procured by the federal government prioritizes truthfulness and ideological neutrality over ideological biases, particularly those associated with diversity, equity, and inclusion (DEI) frameworks.
• It seeks to address perceived distortions in AI outputs caused by ideological agendas, which the order claims undermine accuracy and reliability, citing examples like misrepresentation of historical figures or biased responses.
• Building on Executive Order 13960 (Promoting Trustworthy AI), this order establishes guidelines for federal procurement to avoid AI models that sacrifice factual integrity for preferred outcomes.

Key Actions or Directives

• Establishment of Unbiased AI Principles: Agencies must procure LLMs adhering to two core principles:
• Truth-seeking: LLMs must prioritize historical accuracy, scientific inquiry, and objectivity, acknowledging uncertainty where data is incomplete or contradictory.
• Ideological Neutrality: LLMs must avoid manipulating responses to favor ideological dogmas (e.g., DEI) unless such judgments are transparent and accessible to the user.
• Implementation Guidance: Within 120 days, the Director of the Office of Management and Budget (OMB), in consultation with other officials, must issue guidance on implementing these principles, including:
• Accounting for technical limitations.
• Allowing transparency through disclosure of system prompts or evaluations without requiring sensitive data like model weights.
• Avoiding over-prescription to foster innovation.
• Specifying factors for applying principles to agency-developed LLMs or other AI models.
• Making exceptions for national security systems.
• Agency Responsibilities: Agency heads must:
• Include compliance with Unbiased AI Principles in new federal contracts for LLMs, with vendor liability for decommissioning costs if non-compliant.
• Revise existing contracts where practicable to include these terms.
• Adopt procedures within 90 days of OMB guidance to ensure procured LLMs meet the principles.

Significant Changes to Policy or Law

• Introduces specific Unbiased AI Principles as mandatory criteria for federal procurement of LLMs, marking a shift from broader trustworthy AI guidelines in prior orders (e.g., EO 13960) to a focus on ideological neutrality and truthfulness.
• Explicitly targets DEI-related biases as a primary concern, defining them as including suppression of factual information, manipulation of outputs, and incorporation of specific social theories (e.g., critical race theory, intersectionality), which represents a new policy stance in federal AI procurement.
• Establishes vendor accountability mechanisms, such as bearing decommissioning costs for non-compliance, which could alter contractual relationships in federal procurement.

Potential Impacts

• On Government Agencies: Agencies will need to revise procurement processes and contracts to align with the Unbiased AI Principles, potentially increasing administrative burden and costs in the short term. Exceptions for national security systems may limit the scope of impact in sensitive areas.
• On Citizens: Indirectly, citizens interacting with federal services powered by LLMs may experience outputs shaped by a focus on factual accuracy and neutrality, though the definition of "neutrality" may be contested or perceived differently.
• On Private Sector/Vendors: AI developers and vendors seeking federal contracts must adapt their models or documentation to comply with the principles, which could influence design priorities or increase costs. Transparency requirements may also pose challenges regarding proprietary data.
• On International Relations: While the order focuses on domestic procurement, it could set a precedent for how other nations approach AI bias and regulation, potentially influencing global standards or sparking debate over ideological neutrality in technology.

---

Main Stakeholders Affected

• Federal Agencies: Must implement new procurement guidelines and ensure compliance with Unbiased AI Principles, affecting their operational and contractual strategies.
• AI Developers and Vendors: Face new requirements for transparency and neutrality in LLMs, with financial and design implications for federal contracts.
• Citizens and End Users: May experience changes in the reliability or tone of AI-driven federal services, though the impact is indirect.
• Policy Advocates and Critics of DEI: Groups on both sides of the DEI debate may be affected by the order’s explicit targeting of DEI-related biases, either as a perceived safeguard against ideological overreach or as a restriction on inclusive frameworks.

---

Notable Legal, Constitutional, or Political Implications

Legal Implications

• Contractual Enforcement: The requirement for vendors to bear decommissioning costs for non-compliance introduces a novel accountability mechanism in federal contracts, which could lead to legal disputes over what constitutes "non-compliance" or "reasonable cure periods."
• Scope of Federal Authority: The order limits its scope to procurement rather than private market regulation, likely to avoid legal challenges over federal overreach. However, the broad definition of DEI and ideological bias may invite litigation if vendors or agencies interpret compliance differently.
• No New Rights Created: Section 5 explicitly states the order does not create enforceable rights or benefits, potentially limiting legal challenges from third parties but not precluding disputes within contractual frameworks.

Constitutional Implications

• First Amendment Concerns: While the order focuses on procurement, its emphasis on curbing ideological content (e.g., DEI) in AI outputs could raise questions about free speech, especially if vendors argue that their models’ outputs constitute protected expression. However, as a procurement policy, it may be viewed as a legitimate exercise of government discretion in contracting.
• Executive Authority: The order operates within the President’s authority over federal procurement and executive agencies, as grounded in the Constitution and statutes, likely shielding it from immediate constitutional challenges unless implementation oversteps legal bounds.

Political Implications

• Polarization Over DEI: The explicit targeting of DEI as a source of bias may intensify political debates over the role of social frameworks in technology, positioning the order as a flashpoint in broader cultural and policy conflicts.
• Impact on AI Innovation: By prioritizing neutrality and truthfulness, the order could be seen as either safeguarding AI integrity or stifling diverse perspectives in model design, depending on political viewpoints, potentially influencing future legislative or regulatory approaches to AI.
• Precedent for Future Policy: This order may set a tone for subsequent administrations’ approaches to AI bias, either as a model for stricter controls or a policy to be reversed based on differing ideological priorities.

---

This summary and analysis aim to present the executive order’s content and implications neutrally, focusing on the document’s text and logical outcomes rather than external opinions or biases. If further details or specific aspects require deeper exploration, please let me know.

This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.