Estimated reading time: 5 minutes · Last updated:
An AI-generated image of Bianca Motley Broom, the mayor of College Park, Georgia, circulated widely on Facebook and was later shown to be fabricated. The episode has crystallised a wider public worry: a mid-June poll by Rad Campaign and Women Who Tech found about 71% of respondents worried AI-generated deepfakes and disinformation could affect elections. Where once forgeries took editing skill, modern image generators need only a computer and a prompt, making targeted political misuse cheaper and faster. That gap—between what the public sees online and what can be reliably verified—drives the debate among lawmakers, platforms and advocates over how to limit harm.
I think this is a sad commentary about where we are in our democracy, and where we are in our political discourse,
Bianca Motley Broom
Key takeaways
- Bianca Motley Broom, mayor of College Park, Georgia, was the target of an AI-generated fake image that spread widely on Facebook.
- In a mid-June poll conducted by Rad Campaign and Women Who Tech, about 71% of respondents said they were concerned about AI-generated deepfakes, and roughly two-thirds expressed worry that AI might be used to sexualise women.
- At the federal level, Senator John Kennedy proposed legislation that would obligate AI companies to include a 'kill switch'; Senator Rand Paul blocked the measure, and a Senate amendment to pre-empt state regulation was defeated by a 99-1 vote.
- State approaches differ: Tennessee has enacted the Ensuring Likeness Voice and Image Security Act (the Elvis Act); Georgia advanced deepfake-related bills in 2024 and 2025 but did not pass a law addressing deepfakes.
Table of contents
How easily manipulated images now spread on platforms
Social accounts circulated an image purporting to show Mayor Bianca Motley Broom before independent checks determined it was fake. Where producing convincing altered photos once required advanced editing skills, prompt-driven image generators now let people create lifelike portraits with little technical training. Reuters cited recent examples: a January image posted to the White House account showing Nekima Levy Armstrong, which White House deputy communications director Kaelan Dorr described as a "meme", and AI-produced images of Zohran Mamdami that Reuters identified using Google's SynthID tool.
That combination—a low technical barrier and fast social circulation—means harmful content can reach thousands of users within hours. Political figures such as Senator Raphael Warnock reported AI images portraying him with invented family members or health issues. Those posts are not just embarrassing: as Warnock put it in a Facebook video, fabricated images circulated close to an election risk confusing voters about what is true and what is not.
Lawmakers and states are reacting, but statutes lag the technology
Debate in Congress has shifted from study groups to concrete legislation. Senator John Kennedy introduced a bill that would require AI companies to incorporate a 'kill switch' to disable misbehaving systems; Senator Rand Paul objected and blocked the proposal, calling it premature. Separately, the Senate rejected an amendment intended to prevent states from setting their own AI rules by a 99-1 vote, highlighting bipartisan disagreement over regulatory authority.
State responses are patchy. Tennessee's Ensuring Likeness Voice and Image Security Act, known as the Elvis Act, relies on right-of-publicity law to curb voice cloning and misuse of likenesses. Georgia's legislature advanced bills addressing deepfakes in 2024 and 2025, but those measures did not become law. As a result, mayors and other potential targets face a patchwork of protections and uncertain remedies under existing state defamation and publicity statutes.
Detection tools and the limits of redress
News and technology outlets have begun using technical markers to flag synthetic media: Reuters demonstrated that images of Zohran Mamdami were machine-generated using Google's SynthID. That kind of provenance tool provides a practical countermeasure, but such markers are not applied universally and can be removed or omitted by creators. Platforms frequently depend on third-party fact-checkers and internal moderation, and Reuters documented multiple high-profile cases in which altered images spread before corrections were issued.
Legal remedies remain imperfect. Bianca Motley Broom said she could consider pursuing defamation under Georgia law but acknowledged the need to weigh First Amendment protections. Advocates and legislators quoted by news organisations have called for clearer rules; Rad Campaign and Women Who Tech's polling found roughly two-thirds of respondents were worried that AI could be used to sexualise women, and respondents also raised concerns about AI misleading voters ahead of elections.
| Subject | Platform | Manipulation described | Immediate response |
|---|---|---|---|
| Bianca Motley Broom (College Park mayor) | AI-generated image of her kissing a man | Widely shared then debunked | |
| Nekima Levy Armstrong | X (White House account) | Digitally altered photo (skin darkened, appearing to cry) | White House staff later called it a "meme" |
| Zohran Mamdami | X (parody account) | AI-created images linking him to Jeffrey Epstein | Reuters used SynthID to debunk |
How the situation could move forward
The case for
- Adoption of technical provenance tools such as Google's SynthID and wider newsroom verification practices will make some synthetic content easier to spot.
- Public concern—71% in the cited mid-June poll—gives legislators political cover to pursue clearer limits and disclosure rules.
- State laws like Tennessee's Elvis Act show a legal model that other states could copy to protect likeness and voice.
The case against
- AI image generation is inexpensive and fast, so bad actors can create and distribute new fakes repeatedly.
- Congressional attempts to regulate have met internal resistance: a "kill switch" proposal was blocked by Senator Rand Paul and a pre-emption amendment failed 99-1, showing partisan and procedural hurdles.
- Fragmented state law leaves targets in some jurisdictions—such as Georgia, where no deepfake statute has been enacted despite bills in 2024 and 2025—with limited, uneven remedies.
What to be careful about
- Voter confusion if deepfakes appear shortly before an election and spread faster than verification can occur.
- Gendered harm: two in three respondents in the cited poll worried about AI being used to sexualise women.
- Legal uncertainty for political targets because first amendment limits complicate defamation claims against fabricated images.
- Platform amplification: high-reach accounts can circulate manipulated visuals before provenance tools or fact checks intervene.
The bottom line
The College Park episode is not an isolated stunt but an example of a fast-growing problem: synthetic images that can be created easily and spread widely. Detection tools and state statutes offer partial protection, and the mid-June poll suggests strong public appetite for action. At the same time, federal proposals face procedural resistance and state responses are patchy, leaving a window in which manipulated visuals can influence public opinion. Voters, platforms and lawmakers will need a mix of technical markers, clearer disclosure rules and carefully targeted laws if they are to reduce the risk that what people see online will decisively mislead them.
What to watch
- Watch for any Georgia legislative action on the 2024–2025 deepfake proposals; no date has been set.
- Watch for whether a Senate bill to require technical controls such as a 'kill switch' is reintroduced; no date has been set.
- Watch for wider platform adoption of provenance markers such as SynthID across major social networks; no date has been set.
Frequently asked questions
How common is public worry about AI deepfakes?
The same mid-June Rad Campaign/Women Who Tech poll found that approximately 71% of participants feared AI-generated deepfakes and related disinformation could influence elections, and about two-thirds said they were worried AI could be used to sexualise women.
Are there laws that stop political deepfakes?
Some states offer narrower rules: Tennessee's Ensuring Likeness Voice and Image Security Act (the Elvis Act) limits voice cloning and misuse of likenesses. Georgia considered bills in 2024 and 2025 aimed at deepfakes, but did not enact deepfake-specific legislation.
Can technology reliably detect AI-generated images?
Tools exist and are in use: Reuters applied Google's SynthID to show that certain images were machine-generated. Detection helps, but markers are not universal and can be absent or tampered with, so detection is a partial defence rather than a complete solution.
Related reading
