Trump Administration Mandates Evaluation of AI Bias for Federal Contracts

On December 11, 2025, the Trump administration announced a new directive requiring artificial intelligence (AI) vendors to evaluate and measure political bias in their large language models (LLMs) to qualify for federal agency contracts. This policy applies to all LLMs procured by federal agencies, excluding those used in national security systems.

This directive builds upon President Donald Trump's Executive Order 14319, signed on July 23, 2025, titled "Preventing Woke AI in the Federal Government." The executive order mandates that federal agencies procure only LLMs developed in accordance with two core principles:

1. Truth-seeking: LLMs must provide truthful responses, prioritizing historical accuracy, scientific inquiry, and objectivity, while acknowledging uncertainty when information is incomplete or contradictory.

2. Ideological Neutrality: LLMs should function as neutral, nonpartisan tools, avoiding manipulation of responses in favor of ideological dogmas such as diversity, equity, and inclusion (DEI). Developers are instructed not to intentionally encode partisan or ideological judgments into an LLM's outputs unless prompted by or readily accessible to the end user.

The Office of Management and Budget (OMB) has been tasked with issuing guidance to implement these principles, ensuring that AI systems used by federal agencies adhere to the specified standards.

The policy is expected to significantly impact major technology companies like Microsoft and Meta, which have previously collaborated with the administration to provide AI solutions to the government. Both companies have yet to comment on the new requirements.

In a related action, on December 11, 2025, President Trump signed an executive order aimed at blocking individual states from implementing their own AI regulations. The order argues that diverse state-level rules could hinder innovation and weaken U.S. competitiveness against nations like China, which have centralized AI regulatory systems. The order instructs the Attorney General to challenge state laws and directs the Commerce Department to identify problematic regulations. Additionally, it allows the administration to withhold broadband and other federal grants from states that impose certain AI regulations.

Furthermore, the administration plans to collaborate with Congress to develop a unified national framework for regulating AI. White House adviser Sriram Krishnan emphasized that an executive order signed by President Trump is aimed at streamlining AI innovation by avoiding the complexities of navigating over 1,000 existing state regulations. While the executive order establishes immediate action, Krishnan noted that comprehensive legislation from Congress will ultimately be necessary to create a consistent regulatory environment that supports U.S. leadership in the global AI race.

These developments reflect the administration's ongoing efforts to address concerns about potential biases in AI systems and to ensure that federal AI tools provide accurate and impartial information to the public. By enforcing these standards, the administration seeks to prevent the incorporation of ideological biases into AI outputs, thereby maintaining public trust in government-utilized AI technologies.