As lawmakers begin the 2026 legislative session, an important new phase of the debate over free speech and artificial intelligence is already taking shape in statehouses across the country. Additionally, a new AI bill will govern how people can speak and communicate using machines, raising fundamental constitutional questions about freedom of expression in this country.
The First Amendment applies to artificial intelligence in much the same way as it applied to earlier expressive technologies. Like printing presses, cameras, the internet, and social media, AI is a tool that people use to communicate ideas, access information, and generate knowledge. Regardless of the medium involved, our Constitution protects these forms of expression.
As lawmakers review AI policy in 2026, they are right to reiterate that current law already addresses many of the harms they seek to address (fraud, counterfeiting, defamation, discrimination, election interference), whether or not AI is used. Whether you use a pen or a keyboard, a scam is still a scam. people The person who commits the tort, not the means of committing the tort.
Many of the AI bills introduced or anticipated this year rely on regulatory approaches that raise serious First Amendment concerns. Some require developers and users to attach disclaimers, labels, or other statements to legal AI-generated expressions. This forces developers and users to act as government mouthpieces for views they do not hold. FIRE has long opposed forced speech in schools, campuses, and online, and similar concerns apply when it comes to AI systems.
Election-related deepfake legislation remains a focus in 2026. Over the past year, several states have introduced bills aimed at controlling AI-generated political content. However, these laws often restrict core political speech, and courts have applied well-established First Amendment precedent to find these laws unconstitutional. for example, Coles vs. Bontaa federal district court struck down California’s election-related deepfake law, finding that restrictions on AI-generated political content and associated disclosure requirements violated the First Amendment. The court emphasized that constitutional protections for political speech, including satire, parody, and criticism of public officials, apply even when new technology is used to create that expression.
Another growing category of laws seeks to restrict “chatbots,” or conversational AI, using frameworks borrowed from social media law. These include comprehensive warning requirements that notify users that they are interacting with an AI, and thoroughly enforce many normal, low-risk interactions that do not require warnings. Some proposals would categorically prohibit training chatbots to provide “emotional support” to users, effectively imposing direct and amorphous restrictions on the tone and content of AI-generated responses. Other proposals explicitly or as a practical matter require age or identity verification before users can access the chatbot.
These types of constraints put the government between the people and the information that the people have a constitutionally protected right to access. They censor legitimate expression and burden the right to speak and be heard anonymously.
As such, courts have repeatedly blocked similar restrictions applied to social media users and platforms. Similar results could be achieved with AI.
Broad and comprehensive AI regulation legislation has been introduced again this year, and at least one state has introduced such a proposal so far this cycle. These bills, introduced in multiple states in 2025, go far beyond narrow use cases and seek to impose a sprawling regulatory framework on AI developers, adopters, and users through broad government oversight and comprehensive liability for third-party use of AI tools. Applying these approaches to expressive AI systems raises significant First Amendment concerns, especially when they involve mandatory disclosure and interfere with editorial judgment in AI design.
Addressing real harms such as fraud, discrimination, and election interference can be legitimate legislative goals. But through FIRE’s decades of experience defending free expression, we have observed how broad, vague, and preemptive restrictions on expression tools often chill legitimate speech without effectively targeting illegal activity. This risk is particularly acute when laws incentivize AI developers to suppress legitimate output, limit model functionality, or deny access to information to avoid regulatory risks.
Rather than targeting political speech, imposing age restrictions on tools of expression, or mandating disclosures based on government scripts, government officials should start with the legal tools already available. Existing laws provide remedies for illegal acts and enable enforcement against bad actors without burdening protected expression or innovation. If a gap truly exists, the legislative response must be narrow, precise, and focused on actionable actions.
