You’ve most likely seen at least one of the mailers that sparked the conflict if you’ve been traveling around Montana during this year’s primary season. Political action committee-funded glossy four-color pieces with unflattering pictures of state representatives Sue Vinton and Jennifer Carlson. The faces were recognizable. The clothes weren’t. Neither candidate had ever worn pronoun buttons. Both lawmakers claimed they had never held a Pride banner. Artificial intelligence had been used to create or significantly modify the photographs.
The mailers’ purpose was to disparage the candidates by portraying them in ideological stances that their real records had never endorsed. The ensuing legal dispute, which is currently pending in federal court, may be the nation’s most significant election law battle, but hardly anybody outside of Montana is paying attention to it.
When Montana Senate Bill 25, the legislation at the heart of the lawsuit, was passed, it was the first of its type in the US. Within 60 days of an election, the legislation makes it illegal to utilize artificial intelligence to produce deepfakes of political candidates. However, there is one exception: if the media clearly states that the image was altered, the ban is not applicable. In certain respects, the building is purposefully minimal. Deepfakes are not prohibited. Disclosure is necessary. The current court battle will center on the Helena lawmakers’ framing of the bill as a consumer protection measure rather than a content restriction.
The lead plaintiff in the court challenge, Rep. Alan Bartel, who defended the mailers, has been explicit in his framing. Whether or not the mailers were tasteful is not at issue. It concerns whether Montana can mandate disclaimers on any type of political speech without directly violating the First Amendment.
His lawyer, Matthew Monforton, has claimed that the statute amounts to “speaker-based discrimination” and that the state has established a content-based restriction that the Supreme Court’s free speech jurisprudence consistently rejects by defining a particular category of communication (AI-altered media) and imposing restrictions on it that do not apply to traditional photo editing, satire, or campaign attacks. The underlying mailings’ politics don’t fully capture the complexity of the argument.
Speaking with experts in electoral law, there’s a feeling that this case was bound to occur. Over the past two years, state legislators nationwide have been rushing to pass some form of deepfake election restrictions because to the quick advancement of AI technologies for creating images and videos as well as the justifiable fear that artificial media could skew elections. One of the more comprehensive laws was passed in Montana. Other states have been keeping a tight eye on things. Similar legislation has been passed or proposed in California, Texas, Michigan, and Minnesota. Every other state’s strategy will be influenced by the Montana case, which will result in the first federal court decision on whether this type of rule can withstand constitutional scrutiny.
Carlson and Vinton’s defense is based on a thorough differentiation. They contend that opinions and commentaries are unrestricted by the law. It is completely acceptable for a campaign to claim that Carlson is too sympathetic to progressive causes. Only the unlabeled use of AI to create photos of the candidate doing things she never did is prohibited under the law. The defense claims that this distinction puts the act more in line with conventional consumer fraud and identity theft laws than with prohibitions on speech based on content. The disclaimer requirement is defined as the least restrictive way to remedy a real damage, which is voters being misled about a candidate’s true record.
The courts might be persuaded by such distinction. In certain situations, such as defamation law, commercial fraud, and false assertions made in connection with criminal prosecutions, First Amendment doctrine permits limited regulation of fraudulent speech. The dilemma is whether political imagery produced by AI belongs in that category or if it is securely inside the safe realm of lively, occasionally unpleasant political discourse. This discussion is clouded by the Supreme Court’s 2012 decision in United States v. Alvarez, which invalidated the Stolen Valor Act’s prosecution of making false claims regarding military decorations. The Court’s long-held belief that more speech, not regulation, is the solution to offensive speech still holds true.
The courts may potentially rule in favor of the plaintiffs and invalidate the Act. In 2022, Montana’s prior Clean Campaign Act—which required certain election communications to be transparent—was rejected on First Amendment grounds. In the present instance, the plaintiffs have specifically referenced the decision. The government’s interest in shielding voters from synthetic media does not supersede the speakers’ right to disseminate political communication without state-mandated labeling, according to a convincing case that even disclaimer requirements amount to compelled speech.
As this develops, there is a sense that the case will be determined more on how the federal courts view the larger category of synthetic political media than on the particulars of the Carlson and Vinton mailers. In some respects, the mailers themselves are an uncomfortable collection of facts for both parties. Even sympathetic observers characterize the photos the plaintiffs are defending as deceptive. In addition to more dishonest uses, the defendants are enforcing a statute that prohibits a category of communication that includes overt satire and parody. In order to present their various constitutional arguments, both sides must debate the complexity of the underlying facts.
