Report from NBC Montana
In Brief – US District Judge Susan Watters has ruled that a Montana law requiring disclosures on certain AI-generated political ads likely violates the First Amendment in an order barring the state from enforcing the measure against former state lawmaker Dan Bartel and his political action committee. The law applies to AI-generated ads depicting a candidate or political party with the intent to damage their reputation or deceive voters. Within 60 days of an election, such ads are prohibited unless they carry a disclosure saying the material was made or significantly edited by AI. The law, which imposes civil and criminal penalties, exempts satire and parody, although it does not define those terms. Judge Watters said the statute likely violates the First Amendment because it discriminates based on both content and viewpoint, including that government officials are required to examine the substance of political material and noted that the restrictions target speech considered “injurious rather than uplifting.” Bartel sued after complaints were filed over mailers that included AI-generated images. Although the complaints were dismissed by regulators as satire or parody, Watters nevertheless granted an injunction.
Context – Election “disinformation” and non-consensual pornography are the two most cited potential harms from so-called AI “deepfakes”. More than 25 states have enacted laws against election deepfakes, but federal judges have been very skeptical of their ability to withstand First Amendment scrutiny. None of the other state laws have been enforced against a violator, and the laws in California, Hawaii and now Montana have been blocked in some fashion. On the other hand, Minnesota laws targeting election deepfakes and sexual deepfakes have withstood initial court challenges, although primarily on technical grounds of plaintiffs lacking standing. In Europe, AI “transparency” rules are part of the EU’s AI Act regulatory regime. Corporate requirements include clearly disclosing when users are interacting with AI services and chatbots, or viewing AI-generated or manipulated images, audio and video.