Click play to listen to this article.

(New Hampshire News Service) Newly introduced federal legislation aims to protect voters from the growing threat of political ads generated by artificial intelligence.

Manipulated audio of Vice President Kamala Harris' voice spread widely on social media is just the latest example of a so-called "deepfake" designed to sow confusion ahead of the November election.

Image

Richard Anthony, emerging technologies policy director for the consumer advocacy group Public Citizen, said disruptive AI content in recent European elections revealed the urgent need for action.

"Those incidents were just testing grounds for a big, huge torrent coming our way over the next few months," Anthony predicted.

Anthony explained bipartisan legislation would prohibit the use of AI content for certain political ads and require disclaimers on content substantially generated by AI. Robocalls using AI are already illegal, after a misleading "deepfake" of President Joe Biden before the New Hampshire primary.

Lawmakers say without clear regulations, "deepfakes" could be used to distort election results and widen partisan divides. Earlier this year, tech companies signed a pact to voluntarily prevent AI from disrupting democratic elections but Anthony argued it was largely symbolic and failed to commit companies to banning or removing deceptive content.

He stressed it is important voters be aware of bad actors and if they see a political ad, always check its source.

"Definitely do double-takes on any video, any audio file you see or hear online," Anthony urged.

In the absence of federal rules, eighteen states, including New Hampshire, have already passed laws on AI "deepfakes." Bills are pending in six others.

The Federal Communications Commission has also proposed a rule to require clear disclaimers on AI generated political ads in broadcasting. The public can submit comments though Aug. 12.

Support for this reporting was provided by The Carnegie Corporation of New York.