As the 2025 legislative calendar begins, FIRE is preparing for lawmakers at both the state and federal levels to introduce a deluge of bills targeting artificial intelligence.
The First Amendment applies to artificial intelligence just as it does to other expressive technologies. Like the printing press, the camera, and the internet, AI can be used as an expressive tool — a technological advance that helps us communicate with one another and generate knowledge. As FIRE Executive Vice President Nico Perrino argued in The Los Angeles Times last month: “The Constitution shouldn’t be rewritten for every new communications technology.”
We again remind legislators that existing laws — cabined by the narrow, well-defined exceptions to the First Amendment’s broad protection — already address the vast majority of harms legislatures may seek to counter in the coming year. Laws prohibiting fraud, forgery, discrimination, and defamation, for example, apply regardless of how the unlawful activity is ultimately carried out. Liability for unlawful acts properly falls on the perpetrator of those acts, not the informational or communicative tools they use.
Some legislative initiatives seeking to govern the use of AI raise familiar First Amendment problems. For example, regulatory proposals that would require “watermarks” on artwork created by AI or mandate disclaimers on content generated by AI violate the First Amendment by compelling speech. FIRE has argued against these kinds of efforts to regulate the use of AI, and we will continue to do so — just as we have fought against government attempts to compel speech in school, on campus, or online.
Rather than compelling disclaimers or imposing content-based restrictions on AI-generated expression, legislators should remember the law already protects against defamation, fraud, and other illegal conduct.
Lawmakers have also sought to regulate or even criminalize the use of AI-generated content in election-related communications. But courts have been wary of legislative attempts to control AI’s output when political speech is implicated. Following a First Amendment challenge from a satirist who uses AI to generate parodies of political figures, for example, a federal district court recently enjoined a California statute aimed at “deepfakes” that regulated “materially deceptive” election-related content.
Content-based restrictions like California’s law require strict judicial scrutiny, no matter how the expression is created. As the federal court noted, the constitutional protections “safeguarding the people’s right to criticize government and government officials apply even in the new technological age when media may be digitally altered.” So while lawmakers might harbor “a well-founded fear of a digitally manipulated media landscape,” the court explained, “this fear does not give legislators unbridled license to bulldoze over the longstanding tradition of critique, parody, and satire protected by the First Amendment.”
Artificial intelligence, free speech, and the First Amendment
Issue Pages
FIRE offers an analysis of frequently asked questions about artificial intelligence and its possible implications for free speech and the First Amendment.
Read More
Other legislative proposals threaten the First Amendment by imposing burdens directly on the developers of AI models. In the coming months, for example, Texas lawmakers will consider the Texas Responsible Artificial Intelligence Governance Act, or TRAIGA, a sweeping bill that would impose liability on developers, distributors, and deployers of AI systems that may introduce a risk of “algorithmic discrimination,” including by private actors. The bill vests broad regulatory authority in a newly created state “Artificial Intelligence Council” and imposes steep compliance costs. TRAIGA compels developers to publish regular risk reports, a requirement that will raise First Amendment concerns when applied to an AI model’s expressive output or the use of AI as a tool to facilitate protected expression. Last year, a federal court held a similar reporting requirement imposed on social media platforms was likely unconstitutional.
TRAIGA’s provisions incentivize AI developers to handicap their models to avoid any possibility of offering recommendations that some might deem discriminatory or simply offensive — even if doing so curtails the models’ usefulness or capabilities. Addressing unlawful discrimination is an important legislative aim, and lawmakers are obligated to ensure we all benefit from the equal protection of the law. At the same time, our decades of work defending student and faculty rights has left FIRE all too familiar with the chilling effect on speech that results from expansive or arbitrary interpretations of anti-discrimination law on campus. We will oppose poorly crafted legislative efforts that would functionally build the same chill into artificial intelligence systems.
The sprawling reach of legislative proposals like TRAIGA run headlong into the expressive rights of the people building and using AI models. Rather than compelling disclaimers or imposing content-based restrictions on AI-generated expression, legislators should remember the law already protects against defamation, fraud, and other illegal conduct. And rather than preemptively saddling developers with broad liability for an AI model’s possible output, lawmakers must instead examine the recourse existing laws already provide victims of discrimination against those who would use AI — or any other communicative tool — to unlawful ends.
FIRE will have more to say on the First Amendment threats presented by legislative proposals regarding AI in the weeks and months to come.
