The Executive Order
On 11 December 2025, the President of the United States issued Executive Order 14365, ensuring a National Policy Framework for Artificial Intelligence. The goal of the Executive Order is to remove individual State legal regimes regulating the development and use of artificial intelligence – such as, for example, California’s Transparency in Frontier Artificial Intelligence Act - and to replace it with a “minimally burdensome national policy framework for AI”. The reasons invoked are twofold. First, the Order aims to harmonize AI laws in the United States. Second, the Order claims that some States use AI laws to embed ideology into AI laws, in particular on algorithmic bias.
By itself, the Order does not replace State regulations on AI, as it has not replaced any rules on AI. To achieve this, the Order issues the following instructions to US administrative agencies:
The Artificial Intelligence Litigation Task Force
- Within 30 days of the order (i.e., on 10 January 2026), an Artificial Intelligence Litigation Task Force is created within the Department of Justice, whose function it is to start annulment proceedings with the courts against any and all laws of individual States that violate the policy to be “minimally burdensome”, based on a) their unconstitutional impact on interstate commerce, b) the existence of US federal laws on the matter or c) otherwise being unlawful.
A report on more than “minimally” burdensome AI laws
- Within 90 days of the order, the Secretary of Commerce is to publish a report that identifies “onerous” State laws and laws that should be referred to AI Litigation Task Force. The focus is, in particular, on laws that would require alterations to the “truthful outputs” of the models or that would compel developers or deployers of AI models to disclose information in violation of the First Amendment. This legalese term includes laws that require alteration of training data to avoid algorithmic bias and/or laws that require alterations of the output of language models.
Linking “burdensomeness” of AI laws to access to high-speed Internet subsidies
- The Secretary of Commerce is also to publish a Policy Notice that specifies that States with “onerous” laws are no longer eligible for the Broadband Equity Access and Deployment Program, a subsidy program to fund the construction of high-speed Internet connections. All executive agencies of the US federal government must also assess their grant programs to deny grants to States that have “onerous” AI laws in place and do not commit to removing them or not applying them.
Disclosure on “burdensome” AI laws
- The Chairman of the Federal Communications Commission (the US Telecoms Regulator) must start proceedings to evaluate whether or not to adopt a US standard for disclosure of AI-models that are more than minimally burdensome.
Applying US deceptive advertising rules to AI models
- The Chairman of the Federal Trade Commission shall publish a policy statement on how to apply the US rules on deceptive advertising to AI models. This policy statement must specify the conditions for any alteration of AI model output in relation to it.
Federal Policy Framework
- The Special Advisor on AI and Crypto and the Assistant to the President for Science and Technology have been instructed to jointly prepare a legislative recommendation for a federal policy framework in accordance with the policy of “minimally burdensome” AI laws.
