AN ACT to amend Tennessee Code Annotated, Title 2 and Title 39, Chapter 17, relative to deepfakes in election communications.
SB2321 creates a new Tennessee offense and civil enforcement framework targeting “electoral deepfakes” used in election communications. The bill defines an electoral deepfake as manipulated or AI-generated media that depicts a candidate or political party in a way that would mislead a reasonable person, injure reputation, or create a fundamentally different impression from the original media. It applies only within 90 days of an election and only to depictions of candidates or political parties running for state or local office.
Under the bill, a person, corporation, committee, or other entity may not distribute a prohibited electoral deepfake unless the media includes a clear disclosure that it has been manipulated or generated by artificial intelligence. The disclosure rules vary by format: visual disclosures must be easily readable, video disclosures must remain on screen for the duration, and audio disclosures must be spoken at the beginning and end and periodically in longer recordings. A candidate or political party depicted in a violating deepfake may seek injunctive or other equitable relief, and county election commissions may impose civil penalties ranging from $1,000 to $10,000 depending on the violation history and intent to cause violence or bodily harm.
The bill would amend Tennessee Code Annotated Title 39, Chapter 17 by adding a new section regulating deceptive election-related media, and it also adds a definition to the state’s criminal code for “electoral deepfake.” It creates a new compliance obligation for political advertisers, campaigns, committees, media creators, and other entities distributing AI-generated or manipulated content near an election, while carving out exceptions for bona fide news coverage, certain broadcasters, websites and periodicals that clearly disclaim the content, and satire or parody. It also gives affected candidates and political parties a path to seek court-ordered relief and authorizes county election commissions to assess civil fines.
The bill’s structure suggests a strong concern about election misinformation and the use of artificial intelligence to deceive voters, with the legislation aiming to protect candidates, political parties, and the integrity of election communications. Because no committee transcripts or recorded votes were provided, there is no direct evidence of debate, support, or opposition in the available materials. Based on the text alone, the bill appears to reflect a generally protective and regulatory approach toward AI-generated political content.
The main points of contention likely center on the balance between preventing voter deception and preserving free speech, satire, parody, and press protections. The bill’s broad definitions of manipulated media, its 90-day pre-election restriction, and its penalties for distribution could raise concerns for campaigns, media organizations, and digital platforms about overbreadth or chilling effects. At the same time, the explicit exemptions for news reporting, broadcasters, and parody indicate an attempt to limit those concerns and narrow the law to deceptive election uses.