On Friday, the Trump administration introduced a legislative framework designed to establish a single, national AI policy for the United States. This proposal seeks to consolidate power in Washington by preempting state-specific AI laws, which could undermine the growing initiatives by individual states to regulate the technology’s application and development.
A White House statement accompanying the framework asserted, “This framework’s success hinges on its uniform application throughout the United States. Divergent state laws would impede American innovation and our global leadership in AI.”
The framework details seven main goals, emphasizing AI innovation and expansion, and suggests a federalized strategy to supersede more stringent state regulations. It assigns substantial accountability to parents regarding matters such as child safety, while outlining comparatively flexible and non-binding expectations for platforms.
For instance, it recommends that Congress mandate AI companies implement features to “reduce the risks of sexual exploitation and harm to minors,” yet it refrains from specifying any explicit, enforceable stipulations.
This framework from Trump follows an executive order he signed three months prior, which instructed federal agencies to contest state AI legislation. That order allotted the Commerce Department 90 days to identify “onerous” state AI laws, potentially jeopardizing states’ access to federal funding such as broadband grants. The department has not yet released this list.
Additionally, the order tasked the administration with collaborating with Congress to develop a consistent AI law. This objective is now taking shape, aligning with Trump’s previous AI strategy, which prioritized corporate expansion over regulatory safeguards.
The new framework advocates for a “minimally burdensome national standard,” reflecting the administration’s wider agenda to “remove outdated or unnecessary barriers to innovation” and speed up AI adoption across various sectors. This represents a pro-growth, hands-off regulatory philosophy championed by “accelerationists,” including White House AI czar and venture capitalist David Sacks.
Though the framework acknowledges federalism, the exceptions granted to states are quite limited, maintaining their jurisdiction solely over broad areas such as fraud, child protection, zoning, and state-level AI usage. It firmly opposes states regulating AI development, asserting this to be an “inherently interstate” matter linked to national security and foreign policy.
Furthermore, the framework aims to prevent states from “penaliz[ing] AI developers for a third party’s unlawful conduct involving their models” — a significant liability protection for developers.
Conspicuously absent from this framework are any provisions for liability structures, independent oversight, or enforcement mechanisms to address new harms potentially caused by AI. Essentially, the framework would consolidate AI policymaking in Washington, thereby reducing the capacity for states to serve as initial regulators of nascent risks.
Critics argue that states function as democratic “laboratories” and have historically been more agile in enacting laws concerning emerging threats. For example, New York’s RAISE Act and California’s SB-53 aim to ensure that major AI corporations establish and follow publicly documented safety protocols.
Brendan Steinhauser, CEO of The Alliance for Secure AI, stated, “White House AI czar David Sacks persistently serves the interests of Big Tech, to the detriment of ordinary, hardworking Americans. This federal AI framework intends to bar states from legislating on AI and offers no avenue for AI developers to be held responsible for the damage their products inflict.”
A significant portion of the AI industry welcomes this approach, as it grants them greater freedom to “innovate” without the perceived hindrance of regulation.
Teresa Carlson, president of General Catalyst Institute, informed TechCrunch, “This framework is precisely what startups have sought: a distinct national standard enabling rapid development and scaling. Founders should not be burdened with navigating a disparate array of conflicting state AI laws that stifle innovation.”
Child Safety, Copyright, and Free Expression
The framework’s release coincides with child safety becoming a pivotal issue in the AI discourse. Some states have proactively enacted laws to safeguard minors and impose increased responsibility on technology firms. However, the administration’s proposition diverges, prioritizing parental oversight over platform accountability.
The framework asserts, “Parents are uniquely positioned to manage their children’s digital surroundings and upbringing.” It further states, “The Administration urges Congress to provide parents with effective tools for this purpose, including account controls to safeguard their children’s privacy and supervise their device usage.”
Furthermore, the framework indicates that the administration “believes” AI platforms ought to “implement features to reduce potential sexual exploitation of children and encouragement of self-harm.” While it requests Congress to mandate such protections and confirms that existing laws, including those prohibiting child sexual abuse materials, should extend to AI systems, the proposal utilizes vague terms like “commercially reasonable” and avoids specifying concrete preconditions.
Regarding copyright, the framework endeavors to strike a balance between safeguarding creators and permitting AI systems to be trained on existing content, referencing the principle of “fair use.” This wording echoes the arguments put forth by AI companies amidst a rising number of copyright lawsuits concerning their training data.
The primary safeguards in Trump’s AI framework appear to center on guaranteeing that “AI can pursue truth and accuracy without limitation.” Specifically, its emphasis is on precluding government-initiated censorship, rather than on platform content moderation.
The framework states, “Congress should prohibit the United States government from coercing technology providers, including those in AI, into banning, compelling, or altering content based on partisan or ideological motives.” It further directs Congress to establish a mechanism for Americans to seek legal recourse against government agencies attempting to censor expression on AI platforms or control information generated by them.
This framework emerges while Anthropic is suing the government, alleging a violation of its First Amendment rights after the Department of Defense (DOD) classified it as a supply-chain risk. Anthropic contends that the DOD’s designation is retaliatory because the company restricts the military from using its AI products for mass surveillance of Americans or for autonomous lethal weapon targeting decisions. Trump has described Anthropic and its CEO, Dario Amodei, as “woke” and a “radical leftist.”
The framework’s wording, which prioritizes safeguarding “lawful political expression or dissent,” appears to extend from Trump’s previous executive order aimed at “woke AI,” which encouraged federal agencies to implement ideologically neutral systems.
The distinction between censorship and standard content moderation remains ambiguous, meaning this language could complicate efforts for regulators to collaborate with platforms on concerns such as misinformation, election interference, or public safety hazards.
Samir Jain, vice president of policy at the Center for Democracy and Technology, observed: “While the framework correctly asserts that the government should not compel AI companies to restrict or modify content based on ‘partisan or ideological agendas,’ the Administration’s ‘woke AI’ Executive Order from this summer does precisely that.”