As AI images become common, the threat to elections draws alarm: ‘How will voters know what’s true?’

AI-Generated Images in Politics Prompt Discussions on Verification and Regulation

An image circulated on Facebook showing College Park, Georgia, mayor Bianca Motley Broom kissing a man who was not her husband. The image was later identified as AI-generated, though it had already reached thousands of viewers. Motley Broom stated that the incident reflected challenges in political discourse and reinforced stereotypes about women in leadership positions.

Six years after Motley Broom took office, creating such an image requires only a computer and an AI image generator prompt, compared with earlier needs for image-editing skills. Similar AI-generated images have appeared in other political contexts. In January, the White House posted on X an altered image of Nekima Levy Armstrong that changed her expression and skin tone during an arrest related to a protest. White House deputy communications director Kaelan Dorr described the post as a meme. In February, a parody account on X circulated AI-generated images of New York mayor Zohran Mamdami and his mother with Jeffrey Epstein; Reuters used Google’s SynthID tool to identify them as fabricated.

A June poll by Rad Campaign, a digital marketing firm, and Women Who Tech, an organization focused on female entrepreneurs in technology, found that 71 percent of respondents expressed concern about AI-generated deepfakes and disinformation in elections, while two-thirds reported concern about AI used to sexualize women. The same poll indicated that worry about deepfakes, election disinformation, and sexualization exceeded concern about AI’s effects on jobs or the economy by roughly three to one, according to Rad Campaign founder Allyson Kapin.

Georgia Democratic senator Raphael Warnock posted a video noting AI-generated images that falsely showed him with health issues, nonexistent grandchildren, and a full head of hair. Warnock stated that distinguishing authentic content from fabricated material would grow more difficult and that bad actors could use such tools to divide voters. He described a hypothetical scenario of AI-generated election content released hours before voting and asked how voters would determine accuracy.

Outgoing Louisiana Republican senator John Kennedy introduced legislation requiring AI companies to implement a “kill switch” for certain systems. Kentucky Republican senator Rand Paul blocked the measure, stating that additional input from industry was needed before proceeding. House Democratic leader Hakeem Jeffries called for urgent action on AI rules to establish guardrails against misuse. Jeffries noted that an amendment in the prior year’s omnibus spending bill that would have limited state AI regulation failed in the Senate by a 99-1 vote.

Several states have enacted measures addressing AI use in media. Tennessee’s Ensuring Likeness Voice and Image Security Act protects against unauthorized voice cloning and digital likeness replication through existing right-of-publicity statutes. Most states require clear disclaimers when campaigns employ AI to alter candidate appearances or statements. Georgia has not passed such requirements. Legislation addressing deepfake protections was introduced in Georgia in 2024 and 2025 but has not advanced to enactment.

Motley Broom attributed the image targeting her to political opponents opposed to her prior role as president of the Georgia Municipal Association. She said she would continue her work for College Park and described the images as the work of internet trolls. Georgia defamation law provides a potential avenue for legal response to false images, a remedy Motley Broom indicated she could consider. When asked about legislative responses, she noted that any measures would need to account for First Amendment protections alongside individuals’ interests in avoiding false claims, describing the current situation as an early stage of the issue.

Existing state defamation statutes already allow individuals to seek remedies for demonstrably false images without new AI-specific statutes. Legislative debates continue over the timing and scope of additional rules, with some lawmakers arguing that industry consultation should precede mandates and others calling for prompt action.