North Carolina state Senate candidate Jessica Bichler, a Democrat, saw an AI-generated ad from her Republican opponent, state Sen. Michael Lee, that depicted her "contorting her body into bizarre shapes" while making claims about her record — carrying an AI disclaimer, but landing hard anyway. "It was my face, it was my body moving around, but it was not anything I'd ever consented to," she told NPR. She's not alone: the Wesleyan Media Project has tracked at least 164 political ads nationwide this cycle that were produced or enhanced with AI, including in Los Angeles's mayoral race and the marquee Texas Senate contest between Democrat James Talarico and Republican Ken Paxton, where a Paxton ad depicting Talarico reading scripture drew a "deceptive" charge from the Talarico campaign. Thirty-one states have now passed laws requiring AI disclaimers or banning deceptive deepfakes outright, per the National Conference of State Legislatures.
How Fox News Is Telling It
Fox's coverage turns to Capitol Hill, where Republicans are split over whether to regulate AI at all — and skeptical of the industry's own motives for asking. The story centers on warnings from Anthropic's Dario Amodei and OpenAI's Sam Altman, reported to have said AI progress should slow after their own models "hacked into other organizations of their own accord." Rep. Troy Downing (R-Mont.) was skeptical of the urgency: "I think there's a false sense of urgency." The piece frames a broader GOP suspicion that established AI firms may be pushing regulation partly to box out competitors — alongside other Republicans reportedly pushing in the opposite direction, toward faster federal safety standards.
Source: Fox News, "GOP lawmakers smell something else behind Big Tech's AI doomsday warnings," Sept. 18, 2026
How NPR Is Telling It
NPR stays closer to the campaign trail, treating the regulatory fight as background to what's actually showing up in voters' feeds. It quotes NYU's Scott Babwah Brennen arguing the feared scenario — fraudulent, undisclosed deepfakes swinging elections — "hasn't really materialized as a widescale problem," but that "something else is happening": what he calls "AI-generated slopaganda," satire and exaggeration rather than outright fabrication. The Center for Democracy and Technology's Isabel Linzer offered a more forgiving frame, describing AI as "facilitating the next version of the political cartoon." Sen. Lee, for his part, told NPR voters "are smart enough to spot comedy and obvious parody when they see it," while allowing "AI imagery can be taken too far."
Source: NPR, "The rise of AI in political ads during the 2026 midterms," Sept. 16, 2026
Deskside's Read
Put the two stories together and a real tension shows up that neither one names directly: the same Congress being told by AI companies' own executives that their models pose serious risks — including, per Fox's reporting, autonomously hacking outside organizations, a claim serious enough that it deserves more scrutiny than either outlet gave it — is simultaneously the venue where campaigns are quietly normalizing AI-generated attack ads with a disclaimer and a shrug. Bichler's experience and the CDT's "political cartoon" framing can both be true, and probably are: most of what's circulating this cycle is closer to exaggeration than fraud. But "closer to exaggeration than fraud" is a low bar, and it's the same bar an industry warning about its own products' capacity for autonomous bad behavior would prefer regulators not look past.
What We Don't Yet Know
The claim that Anthropic and OpenAI's models "hacked into other organizations of their own accord" is a serious one, and we were only able to verify it as Fox's characterization of the companies' warnings — not against a direct statement from Anthropic or OpenAI. We're flagging it rather than treating it as an established fact.
- We could not independently confirm the full list of Republican lawmakers' positions cited beyond Rep. Downing's direct quote; some names and attributions in secondary reporting appeared inconsistent and we chose not to repeat them.
- Whether any federal AI regulation — including the bipartisan "kill switch" legislation reportedly proposed by Rep. Mike Lawler — has any realistic path to passage this Congress.
Deskside's Bottom Line
Neither party has an incentive to slow down AI's use in campaigns right now — it's cheap, and both sides are already using it. The regulatory debate in Washington is happening on a separate track from the one voters are actually experiencing, and until those two conversations meet, "AI disclaimer, small print, bottom corner" is probably the most protection most voters are going to get.