'This Should Be Illegal': Senate GOP Uses AI Deepfake to Attack Talarico

In the latest example of Republicans using artificially generated deepfakes to attack their opponents, the Senate GOP’s official social media account has posted an attack ad depicting a synthetic version of Texas Democrat James Talarico, a state representative and US Senate candidate.

The video, posted on Wednesday to the National Republican Senatorial Committee (NRSC) page on X, portrays a frighteningly realistic approximation of Talarico's (D-50) appearance and voice.

The state representative, who won the Democratic nomination for Texas’ US Senate seat in a primary earlier this month, is depicted reading an array of old social media posts that the NRSC described as “extreme statements praising transgenderism, twisting Christian beliefs, and advocating for open borders.”

The posts were all real. Talarico did indeed state, following a spate of mass shootings against minorities in 2021, that "radicalized white men are the greatest domestic terrorist threat in our country." He also did say that his office had added personal pronouns to official business cards out of respect for transgender Texans, that he believed God was "nonbinary," and that he was "the only teenage boy at Planned Parenthood's March for Women's Lives in 2004."

However, all of the posts are at least several years—if not more than a decade—old. The video also depicts its AI simulacrum of Talarico smiling and reminiscing fondly about the posts, which he never actually did.

"So true," he is depicted saying after reading the tweet about "radicalized white men." "I love this one too," he says before reading the post about "pronouns."

Aside from a small, translucent watermark in the bottom-right corner of the video, labeling it "AI Generated," there is no indication that the video is a fabrication.

While both sides of the aisle have dabbled in the use of AI to attack their opponents, Politico's Adam Wren has noted that deepfakes were not being deployed equally and have become central to the "approach" of the GOP in campaigns.

In October, after Republicans made a similar video showing a simulated Senate Minority Leader Chuck Schumer (D-NY) celebrating the government shutdown, Wren noted the frequency with which such tactics were being used by Republican campaigns at both the state and federal level:

Other examples of AI-generated advertising have also come from Republicans. An ad for Mike Braun, now governor of Indiana, last year used AI to fake scenes, without disclosing it. President Donald Trump’s account regularly posts clearly fake videos of the president ridiculing opponents...

The [NRSC] released one hitting Democratic Maine Gov. Janet Mills as she launched her Senate campaign, and one simulating a Democratic group chat.

Deepfakes have also been deployed heavily by social media accounts for President Donald Trump's White House to degrade opponents.

Earlier this year, the official account posted a photo of an organizer who’d been arrested during a protest against US Immigration and Customs Enforcement (ICE), doctored to portray her uncontrollably crying, when actual photos of the event show her appearing stone-faced and stoic while being led away in handcuffs.

While more than half of all US states have legislation regulating the use of AI deepfakes for election-related content, the consumer advocacy group Public Citizen has said such content needs to be addressed at the federal level.

The group has called on the Federal Elections Commission (FEC) to designate the use of AI for deceptive political messaging as fraudulent misrepresentation and on Congress to pass legislation banning the practice and requiring AI-generated content to be prominently labeled.

Robert Weissman, the co-president of Public Citizen, told Common Dreams that the deepfake of Talarico "is a disgrace and the NRSC should put it down immediately."

"Political deepfakes are a profound threat to our democracy, because there is no realistic way for voters to understand they are seeing fake representations rather than real video," Weissman said. "This deepfake has an 'AI-generated' watermark, but it’s all but invisible–sort of like an admission of wrongdoing, more than an effort at transparency.”