Political campaigns will need to disclose deepfakes
Add Axios as your preferred source to
see more of our stories on Google.

Illustration: Natalie Peeples/Axios
A new Louisiana law went into effect Aug. 1 that requires disclosures when artificial intelligence is used in political ads to depict candidates for office.
Why it matters: AI-manipulated campaign content was already an issue in the recent Senate primary.
How it works: The new law targets deepfakes, or AI-generated depictions of real people doing things the subject may have never actually done.
- If used in political ads and not disclosed, according to the new law, campaigns could face up to $2,000 in fines.
The big picture: More than 30 states have already passed restrictions on the use of AI in campaign materials, according to NPR.
- But recent research shows that simply disclosing its use may not be enough to clarify what's actually happening in campaign ads, says Nihar Sreepada, the co-director of the Manship School of Mass Communication's AI and Misinformation Lab at LSU.
- "Transparency by itself is insufficient to entirely negate the influence of deepfake videos," he says, citing a study where more than half of participants still believed a deepfake video despite being warned in advance that it was AI-generated.
Plus, disclosures can have a side effect of making other content more believable because it doesn't have the labeling, Sreepada says.
Zoom in: In the Republican Senate primary early this summer, campaign ads depicted deepfakes of Rep. Julia Letlow and State Treasurer John Fleming.
- Fleming reshared on social media one such video depicting Letlow referencing her late husband, according to Fox 8. He later deleted the post.
- In another video, Fleming was shown driving a bus of immigrants. It "ran constantly," recalls Fleming, who chalks up the use of AI to tech-savvy parody.
Between the lines: Fleming, who lost the race to Letlow, doesn't believe AI in ads is a problem. The issue, he says, is restating false information.
- "You can say dishonest things whether they're AI or not," he says.
The bottom line: Rep. Mandie Landry, who authored the new legislation, anticipates tightening AI regulations next session, according to the Louisiana Radio Network.
- Either way, "no solution is completely perfect in a world where technology is developing so quickly," Sreepada says, making AI literacy an ever more important skill.
