Google has announced that from mid-November 2023, political advertisers must prominently disclose if their advertisements feature artificial intelligence content showcasing real or seemingly real individuals or events. This revision is integrated into Google's Political content policy. The disclosure should be evident and positioned where viewers can easily spot it. The policy will encompass various content types such as images, videos, and audio.
Specifics of the Policy
Advertisements containing AI content modified in ways that don't significantly impact the ad's message will not need this disclosure. Such minor modifications encompass image adjustments like resizing, cropping, color enhancements, defect fixes such as “red eye” removal, or background changes that don't fabricate realistic representations of genuine events.
On the other hand, advertisements that require a distinct and evident disclosure are those with AI content portraying someone uttering or performing actions they didn't genuinely execute. Moreover, ads that modify real event footage or fabricate a genuine-looking event representation to showcase scenes that never occurred will also mandate the disclosure.
Background and Context
AI-driven images and soundtracks have gained traction during the 2024 Presidential campaign. Google's move is a response to growing apprehensions about disinformation proliferation. The company's updated regulations derive from its pre-existing Manipulated Media Policy and are slated for activation in November. Allie Bodack, a representative from Google, told The Verge, “Given the escalating use of synthetic content tools, we've decided to broaden our guidelines, compelling advertisers to reveal if their election ads encompass digitally modified or produced material”.
Contrastingly, other social media giants have taken different stances. For instance, X/Twitter recently pivoted from its earlier position, now permitting political ads. Concurrently, Meta has been under scrutiny for its ad moderation practices.
AI in Political Campaigns
Numerous political campaigns have harnessed AI for their promotional content. The Republican National Committee, for instance, unveiled an ad with AI-produced visuals critiquing President Joe Biden's reelection campaign. Florida's Governor, Ron DeSantis, employed AI-produced visuals of Donald Trump and Anthony Fauci, the ex-chief medical advisor to the White House, in a critical advertisement.
TrumporBiden2024 is a an AI power debate between Donald Trump and Joe Biden that was running on Twitch constantly. As the days went by the AI became increasingly nonsensical.
The AI versions of Biden and Trump are not only capable of mimicking their voices, but also their personalities and quirks. However, sometimes they go off the rails and say things that are completely nonsensical or hilarious.
During the debate the AI candidates followed each other's cues and acted as if they were having a serious debate, complete with facial expressions and gestures. But the absurdity of their arguments made it clear that they were not real.
Such AI-driven advertisements have sparked debates among legislators. Yvette Clarke, a representative from New York, proposed legislation necessitating disclosures for political ads featuring AI-produced content. The Federal Election Commission is also mulling over limitations on election advertisements employing AI.