The video released by the National Republican Senatorial Committee shows an AI-generated version of Democratic Senate Minority Leader Chuck Schumer repeatedly saying “Every day gets better for us” and smiling.
@NRSC/Annotation by NPR
hide caption
toggle caption
@NRSC/Annotation by NPR
A new attack ad from Senate Republicans uses Sen. Chuck Schumer’s real words about shutting down the government — but in an AI deepfake of the Democratic Senate Minority Leader.
The 30-second video posted Friday to X and YouTube by the National Republican Senatorial Committee alarmed many observers who warned it crossed a new political boundary and could trigger a flood of AI-generated deepfake attack ads.
Captioned on The video ends with Schumer smiling as a narrator says, “Shutting down Schumer is making things worse across America and Democrats love it.”
“Every day gets better for us” was part of a quote Schumer gave to the political site Punchbowl News in an Oct. 9 article about the shutdown. He continued: “That’s because we thought about it well in advance and we knew health care was going to be the focus on September 30 and we prepared for it… Their whole theory was: Threaten us, bamboozle us, and we would submit in a day or two.”
However, the original story only included the text of Schumer’s statement, no video or audio.
An NRSC press release announcing the ad describes it as “visualizing” Schumer’s comment and says it was “created in part using artificial intelligence software and includes an AI disclaimer.”
Schumer’s fake images include a small transparent watermark of the NRSC logo and the words “AI GENERATED” in the lower right corner. On YouTube, the video has an additional label “Edited or Synthetic Content” stating “The audio or visuals have been significantly altered or digitally generated.”
“These are Chuck Schumer’s own words,” Joanna Rodriguez, NRSC communications director, said in an email to NPR. “We know Schumer would like people not to know he said that, but he did, and this video is a way for voters to see and hear how proud he is of his closure.”
Schumer’s office did not respond to a request for comment.
Even if Schumer actually said that quote, using AI to invent fake video footage crosses the line, said Hany Farid, a professor at the University of California, Berkeley who studies manipulated media.
“If they wanted to use the quote, they could have shown an image of Schumer and superimposed the quote, as is often done. One cannot reasonably argue that creating a deepfake of Senator Schumer is the same thing,” he said.
Additionally, Farid said the disclaimer in the bottom right corner of the video “is not a particularly clear or obvious disclosure, especially when most users are stuck scrolling through their social media posts.”
The NRSC video was also criticized by journalists and other commentators on social media, including journalists from right-wing publications.
“The Schumer piece is an AI-generated video — which is leaked in the bottom right corner — but we are on the threshold of a terrifying new world,” Jon Levine, a political reporter at the Washington Free Beacon, wrote on X.
“The NRSC is increasingly relying on AI for its ads, but this is the first time I’ve seen them go so far as to use it for a fake video of an opponent. It’s a slippery slope, even with the small disclosure of AI in the corner,” Washington Examiner Congressional Reporter Ramsey Touchberry posted on X.
Amid the criticism, NRSC’s Rodriguez wrote on his own X account: “The AI is here and not going anywhere. Adapt and win or catch pearls and lose.”
This isn’t the first time Republicans have used AI to impersonate Schumer and other Democrats. On the eve of the shutdown, President Trump posted an apparently AI-generated, profanity-filled video to Truth Social, inventing Schumer saying that “no one likes Democrats anymore.” House Minority Leader Hakeem Jeffries is also shown in the video wearing a cartoonish sombrero and mustache.
But unlike Schumer’s NRSC video, earlier examples were easy to spot as inauthentic.
Farid warned that the growing prevalence of AI fakes in politics could have a corrosive effect.
“While I don’t think our leaders should post misleading deepfakes, they also run the risk that when they post real content, the public will have good reason to question its authenticity,” he said.
