Artificial intelligence; prohibiting distribution of certain media and requiring certain disclosures. Effective date.
SB 894 creates a new Oklahoma law aimed at election-related deepfakes and other synthetic media generated or altered with artificial intelligence. The bill defines “artificial intelligence,” “synthetic media,” and “deepfake,” with the deepfake definition focused on deceptive media depicting a candidate or political party in a way that could mislead viewers or damage reputation. It generally prohibits a person, corporation, committee, or other entity from creating or originally distributing such synthetic media within 90 days of an election if they know it is a deepfake of a candidate or political party appearing on the ballot.
The bill allows the prohibition to be avoided if the media includes a required disclosure stating that the image, audio, or video has been manipulated or generated by artificial intelligence. It specifies how disclosures must appear for visual and audio media, including font size and timing requirements, and gives candidates depicted in deepfakes the right to seek injunctive relief, damages, court costs, and attorney fees. The State Election Board may also impose civil penalties, with higher fines for repeat violations or conduct intended to cause violence or bodily harm. The bill includes exceptions for bona fide news coverage, certain paid broadcasts, legally required candidate messages, general-interest publications that clearly acknowledge authenticity concerns, and satire or parody.
If enacted, SB 894 would add a new election-integrity and media-disclosure provision to Title 75A of the Oklahoma Statutes, regulating the use of AI-generated or AI-manipulated media in the 90 days before an election. It would create new compliance obligations for political actors, media distributors, and others who create or disseminate synthetic media involving candidates or political parties, while also giving candidates and the State Election Board enforcement tools through civil litigation and administrative penalties.
The available voting history suggests the bill had at least some support in the Senate committee, passing 5-2 on a DO PASS vote. The bill’s structure and exceptions indicate an effort to balance concerns about deceptive AI-generated election content with protections for news reporting, satire, and other legitimate speech. No committee transcript is available, so the record does not show detailed debate, but the vote suggests the measure was viewed favorably by a majority while still drawing some opposition.
The main points of contention are likely to be the scope of the ban, the 90-day pre-election window, and the potential impact on political speech and media use. Critics may argue that the definitions of deepfake and synthetic media could be broad or difficult to apply, and that civil penalties and private rights of action could chill legitimate expression. Supporters are likely to emphasize preventing deceptive AI-generated election interference, protecting candidates from reputational harm, and requiring clear disclosures so voters can better evaluate manipulated media.