AI Regulation Tracker / Courts and free speech
A Federal Court Just Struck Down Hawaii's Election Deepfake Law
On January 30, 2026, a federal judge permanently blocked Hawaii's Act 191, the state's attempt to police AI deepfakes in elections. The ruling is a warning label for every executive and policy team building deepfake or synthetic-media rules: content-based speech restrictions collide with the First Amendment, and courts are enforcing that line.
Hawaii passed Act 191 in 2024 to do something a lot of states have tried: keep AI-generated deepfakes from poisoning elections. The law prohibited anyone from recklessly distributing "materially deceptive media" involving candidates and elected officials during the election season, roughly February through November, with a safe harbor if you attached a qualifying disclaimer and carve-outs for broadcasters and service providers. On paper it looks like reasonable guardrails against a real problem. On January 30, 2026, a federal court held that it violates the First Amendment and permanently enjoined the state from enforcing it.
The plaintiff tells you a lot about why. The Babylon Bee, a satire publisher, sued alongside a Hawaii resident, arguing the law punished exactly the kind of exaggerated, obviously-not-literal political content that the First Amendment has always protected. Satire lives on depicting public figures saying things they never said. A law that makes "materially deceptive media" about candidates a legal risk sweeps that in, and the court said so.
What did the court actually rule?
Judge Shanlyn A.S. Park found Act 191 "presumptively invalid" because it discriminates based on content and speaker, "and, in doing so, restricts constitutionally protected political speech." That is the core problem. The law only reached electoral communications, which means it singled out speech by its subject matter, and content-based restrictions on political speech get the most demanding review a court can apply. The judge also found the law did not fit any historical First Amendment exception like fraud or defamation, and that its prohibition on speech carrying a "risk of harming" candidates or officials was too broad to survive.
The court did not stop at the First Amendment. It also held the law unconstitutionally vague under the Fourteenth Amendment, describing the standard it set as "an inherently subjective assessment for enforcement agencies." When a statute leaves regulators guessing about what counts as a violation, it invites arbitrary enforcement, and that alone can sink it. Two independent constitutional defects, one order, permanent injunction.
Why should executives outside Hawaii care?
Because this is the pattern, not the exception. States have been racing to pass election-deepfake laws, and the ones written as content-based bans keep running into the same First Amendment wall. California's AB 2839 hit trouble on similar grounds. What the Hawaii ruling adds is a clean, recent, permanent decision laying out exactly why the ban model fails: it targets speech by topic, it lacks a traditional unprotected-speech hook, and it is vague about enforcement. Any company relying on a patchwork of state deepfake laws to define its own obligations is relying on ground that is actively shifting under litigation.
There is a design lesson buried in the reasoning, and it is the useful part. Courts have been far more receptive to disclosure and labeling requirements than to outright bans on categories of content. A rule that says "if this is synthetic, label it" regulates the manner of speech. A rule that says "you may not distribute deceptive media about candidates" regulates the content of speech, and that is the one that draws strict scrutiny. If you are building synthetic-media policy, that distinction is the whole game.
What should policy and legal teams do now?
If your company writes platform rules, model-safety commitments, or internal synthetic-media policy, anchor them in provenance and disclosure, things like content credentials, clear AI-generated labels, and takedown processes tied to impersonation or fraud, rather than in bans on political subject matter that a court can call content discrimination. If you advise campaigns, communicators, or media clients, understand that in Hawaii Act 191 is now dead and that similar state statutes elsewhere are vulnerable, so do not over-rely on them. And watch the appeal path. A permanent injunction from a district court can be taken to the Ninth Circuit, so treat this as strong current law in Hawaii and a strong signal nationally, while tracking whether the state appeals.
Questions professionals are asking
Is Hawaii's Act 191 still in effect?
No. On January 30, 2026, the U.S. District Court for the District of Hawaii entered a permanent injunction barring the state from enforcing Act 191, and declared it unconstitutional. The state could appeal to the Ninth Circuit, but as of now the law cannot be enforced.
Why did the court strike it down?
The court held the law "presumptively invalid" because it discriminates based on content and speaker and restricts protected political speech, and it did not fit any traditional unprotected-speech category. It separately found the law unconstitutionally vague, calling its standard "an inherently subjective assessment for enforcement agencies."
Does this mean all deepfake election laws are unconstitutional?
Not automatically, but content-based bans are in serious trouble. Courts apply strict scrutiny to laws that target political speech by subject matter, and several have fallen. Disclosure and labeling requirements, which regulate the manner rather than the content of speech, have fared better.
What should our synthetic-media policy look like instead?
Anchor it in provenance and disclosure. Use content credentials, clear AI-generated labels, and takedown processes tied to impersonation or fraud, rather than bans on political subject matter. That approach regulates how content is presented, which courts have been far more willing to uphold than bans on what content may say.
RELATED BRIEFINGS
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any statute or court ruling applies to your situation with qualified counsel in the relevant jurisdiction.