Revise election laws regarding disclosure requirements for the use of AI in elections
SB 25 creates a new set of election-law rules aimed at deepfakes and other AI-generated or synthetic media used in election communications and electioneering communications. It defines terms such as “AI-generated content,” “deepfake,” “generative artificial intelligence,” and “synthetic media,” and applies the restrictions to content involving candidates, including incumbents and current officeholders, as well as political parties.
The bill generally prohibits distributing, publishing, broadcasting, transmitting, displaying, or paying for a deepfake of a candidate or political party within 60 days of the start of voting in an election, unless the communication includes a specified disclosure stating that the image/audio/video/multimedia has been significantly edited by artificial intelligence and may falsely appear authentic. It also sets formatting and timing requirements for disclosures in print, video, internet, and audio formats, and creates exceptions for bona fide news coverage, certain paid broadcast situations, satire or parody with disclaimer, distribution platforms that provide required disclaimers, and certain internet/service providers for third-party content. The bill is intended to be codified into Title 13, chapter 35 of Montana law.
SB 25 adds a new part to Montana’s election code regulating AI-generated political content and creates both civil and criminal enforcement mechanisms. It authorizes injunctive relief, attorney fees, actual damages, and up to $10,000 in punitive damages in civil actions brought by an aggrieved candidate or political party, and it allows complaints to be filed with the Commissioner of Political Practices, with escalating penalties that can include civil penalties, misdemeanor prosecution, and felony prosecution for repeated sufficiency findings. The bill therefore affects candidates, political parties, broadcasters, online platforms, and other entities involved in election communications, while also shaping how AI-generated political ads and media must be labeled or withheld during the pre-election period.
The overall sentiment reflected in the voting history is broadly supportive, with the bill passing both chambers and ultimately being enacted. It received strong majorities in the Senate and House on final passage, though not without some opposition, especially during House consideration. The absence of committee transcript excerpts limits direct insight into debate, but the vote margins suggest general bipartisan concern about deceptive AI use in elections alongside a meaningful minority skeptical of the bill’s approach or scope.
The main points of contention appear to center on how far the state should go in regulating political speech that uses AI, and whether the disclosure mandate and penalties are too broad or burdensome. The failed House motion to amend indicates there was at least some effort to change the bill’s terms, likely around scope, enforcement, or exemptions. Likely areas of dispute include the 60-day pre-election restriction, the definition of a deepfake, the treatment of satire and parody, and the potential liability for broadcasters, platforms, and service providers that carry third-party content.