Washington intensifies the hunt for laws that run counter to its AI doctrine.
Its AI Action Plan released this summer paved the way. It notably called for federal agencies funding AI programs to weigh the regulatory climate in the states. And to curb such funding where a regime could compromise their effectiveness.
The White House pledged more broadly to review or repeal any text that unnecessarily hinders AI deployment. Front and center were final orders, executive decrees, and FTC injunctions (the Federal Trade Commission, which enforces consumer law).
The Ideological Bias Mantra
The operating approach became clearer with an executive order signed last week by Donald Trump. In its crosshairs are laws that “require the integration of ideological biases into models.” Reference is made to a “Colorado law.” It is probably SB24-205 (“Consumer Protections for Artificial Intelligence”), which is due to come into effect on February 1, 2026. The premise: by banning “algorithmic discrimination,” the text could force models to produce false outputs to avoid a “differential treatment” of minorities.
The executive order also targets laws with extraterritorial reach that would, in turn, jeopardize interstate commerce.
A Government Task Force to Challenge Texts Believed Illegal
This pruning work is meant to “maintain and strengthen the United States’ global lead in AI through a national framework that is as painless as possible.”
A working group will be charged with challenging (challenge) laws that conflict with this objective. The Attorney General has 30 days—from the signing of the executive order—to establish it.
This challenge may proceed, notably, on grounds of unconstitutionality, the primacy of federal law… or “any presumption of illegality on the part of the Attorney General.”
The Prospect of Cutting Federal Funding
The Secretary of Commerce has 90 days to identify the problematic laws and to flag them for the task force. He will be required to list at least those that “require AI models to alter their outputs to be truthful” or that could compel developers or deployers to disclose information in contravention of the Constitution, starting with the First Amendment (freedom of speech).
Within the same period, the Secretary must issue a notice regarding BEAD funding (Broadband Equity Access and Deployment, the federal program worth $42.5 billion). Funds not allocated to the program’s core infrastructure deployment goals would not be accessible to states enacting or considering enacting problematic laws. In parallel, executive departments and agencies are invited to review their grant programs to determine if they can condition them on the absence of these laws—or on a binding commitment not to enforce them.
Facing Altered Outputs, Washington Advances… Consumer Law
The FCC (Federal Communications Commission) has 90 days to initiate a proceeding to determine whether a federal standard for disclosing information related to AI models should be adopted. It would prevail over state laws.
Within the same timeframe, the FTC must issue a principles statement on applying consumer protection law to AI. It will need to spell out more precisely the circumstances in which rules forcing altered outputs do not take precedence over federal law prohibiting unfair or deceptive practices.
Ultimately, there would be a recommendation for a uniform federal legislative framework. That framework would prevail over the problematic AI regulations. But not, however, over laws addressing child protection, computing and data infrastructure, and public procurement.
