I have three things to say about this.
Facing increasingly tight polling and an apparent inability to make dirt stick on their Democratic rivals, two of Texas’ highest-profile Republican elected officials have resorted to using artificial intelligence to generate phony footage for their attack ads.
Lt. Gov. Patrick used controversial AI deepfake technology in a spot that debuted over Labor Day weekend mocking Texas Democrats running this election cycle.
Meanwhile, Attorney General Ken Paxton’s campaign debuted an AI-generated ad showing James Talarico reading fake passages from the Bible — an apparent attempt to slam the state rep’s background as a seminarian. The spot was released to coincided with the Republican Midterm Convention in Dallas.
“Ken Paxton and his billionaire backers are using AI deepfake videos to literally put words in James’ mouth because they can’t rely on facts to win this race,” Talarico campaign spokesman JT Ennis told the Current.
Patrick’s political ad, posted on his social media accounts, featured the likenesses of his Democratic challenger Vikki Goodwin as well as those of Talarico and gubernatorial candidate Gina Hinojosa, also Democrats.
The phony clip shows Goodwin playing spoilsport at a University of Texas Longhorns football game as she proposes new taxes, including a soda tax, a brisket tax, a happiness tax and a “nostalgia surcharge.”
“They should be chanting taxes instead!” the digitally generated Goodwin is shown saying as the crowd chants “Texas.”
Next, the clip shows a phony Gina Hinojosa appearing at Texas Tech to demand a permit for Red Raider fans’ finger guns.
Finally, the spot includes fake footage of Talarico at a Texas A&M game eating tofu and asking why the 12th Man has to be so gendered.
Goodwin told the Current during a Wednesday phone call that the ads’ fake images and fabricated statements show the desperation of Patrick’s campaign.
“My platform is something that resonates with people, so I don’t really need to lie,” Goodwin said. “There’s a lot of information out there about my opponent already, so there’s not any need for me to put out negative AI ads.”
1. This is both desperate, as Rep. Goodwin says, and pathetic. I will be far from the first person to observe how these guys never had to engage with their opponents before, and it shows. We can all have our own opinions of AI and its usage, but we can all also agree that chumps make AI slop like this. You can see the ads embedded in the story, but why would you? Your brain gets rotted enough by the Internet without going out of your way to watch this stuff.
2. That doesn’t mean it won’t work. People aren’t good at all at distinguishing deepfakes from reality, mostly because at this point the video quality of the deepfakes is too damn convincing. There are other context clues, and really anyone with an ounce of critical thinking ability ought to see right through these. But it might sway some reluctant Republicans to get out and vote. Whatever happens in the election, we’ll never really know what if any effect they had. I doubt it will be much, but I also doubt it will be zero.
3. There’s no valid reason why these ads should be allowed to run. Spare me your First Amendment arguments, this is “shouting FIRE in a crowded theater” stuff to me. The public good outweighs Dan Patrick’s artistic license. I have no idea what the best way to disincentivize the use of these things, but I’m open to any and all reasonable suggestions. It will be nothing but this from now till the end of time if there’s any belief that they’re effective. Those who will want to outlaw this crap will have a very limited window of opportunity to take action against it.
UPDATE: Per John Moritz in the Chron:
All this begs the question: Is it legal to use deepfake ads in political races in Texas?
Short answer, yes. But a 2019 state law says they cannot be used in the final 30 days of a campaign. Deepfakes are not regulated in federal races.
Good to know. Let’s all be prepared to file complaints if we see them after October 4.
I think that everyone should be granted NIL rights concerning AI. No one should be allowed to use computer generated images of you (or representations of your voice) without your permission.
Agreed! Deepfakes seem like straight forward libel.