People have been fretting about online lies influencing elections for a very long time—and for good reason. There may be little evidence of “fake news” actually swaying masses of voters, but it certainly fools some people. And most of us increasingly get our news from social media, where misinformation can spread around the world in seconds.
Add AI deepfakes to the mix and—especially since generative AI got good in the last couple years—it again makes sense that people are freaking out. Except that, again, there’s very little evidence of deepfakes actually changing many people’s minds.
Well, now we have reason to believe that AI deepfakes can influence elections—just not in the way that their distributors were hoping.
Last night, after the first (and probably last) televised debate between presidential hopefuls Kamala Harris and Donald Trump, Taylor Swift endorsed Harris. This is in itself not a surprise; she endorsed the Biden-Harris ticket in 2020. But, four years on, Swift has developed a unique level of superstardom that gives her voice the weight of Tom Hanks, Dolly Parton, and the Pope all rolled into one. And until Tuesday night, she had refrained from using that capital to steer her fans one way or another in this momentous election.
As Swift made clear in her Instagram statement, she is endorsing Harris and running mate Tim Walz because she personally agrees with their demeanor and policies. “Your research is all yours to do, and the choice is yours to make,” she insisted to her fans. But she also left no doubt that Trump had largely triggered her decision to go public with her opinion, by disseminating AI deepfakes of her and her fans last month.
The deepfakes in question were deeply silly. The one that featured Swift herself was an obviously bogus riff on the iconic, finger-pointing U.S. Army recruitment poster from the First World War, saying: “Taylor wants you to vote for Donald Trump.” Further images claimed to depict crowds of “Swifties for Trump,” though all the faces had the uncanny-valley glossy sheen that (for now) continues to betray the AI origins of faked photos. “I accept!” Wrote Trump in the Truth Social post that he used to spread the images.
Tech-savvy viewers would not be fooled, and it’s quite possible that Trump was sharing the images (which neither he nor his team created) as a joke. But Swift clearly thought some people would be taken in. And—being the world’s premier revenge expert—she saved her payback for the most impactful moment.
“Recently I was made aware that AI of ‘me’ falsely endorsing Donald Trump’s presidential run was posted to his site,” she wrote yesterday. “It really conjured up my fears around AI, and the dangers of spreading misinformation. It brought me to the conclusion that I need to be very transparent about my actual plans for this election as a voter. The simplest way to combat misinformation is with the truth.”
With the election being as tight as it is—and with eight weeks’ worth of potentially course-changing events still to come—it would be foolish to predict that Swift’s intervention will be decisive. But, coming straight after a debate that Harris is widely seen as having won, and given Swift’s exceptional influence, it’s certainly a major boost for the vice president. And without AI, it probably wouldn’t have happened.
More news below.
David Meyer
Want to send thoughts or suggestions to Data Sheet? Drop a line here.
NEWSWORTHY
New PlayStation. Sony has officially unveiled the PlayStation 5 Pro, the latest version of the fifth generation of its gaming console, which first appeared in 2020. As IGN explains, the new PlayStation costs $699 and does not come with a disc drive, though one can be bought as an optional extra. As one would expect with the “Pro” branding, it is significantly more powerful than its predecessors—and a couple hundred bucks more expensive.
Utah social media law. Utah’s Minor Protection in Social Media Act was supposed to go into effect at the start of next month, but a federal judge has blocked that from happening. The law requires social media companies to give kids the maximum privacy settings by default, restrict the visibility of their accounts, and verify the ages of their users. As Deseret News reports, Chief Judge Robert J. Shelby said NetChoice (the tech industry group that asked for the injunction) had made a “persuasive” argument that the law violates the First Amendment by telling social media firms how to construct and operate their platforms.
Starship delay frustration. Elon Musk’s SpaceX is furious at the Federal Aviation Administration’s decision not to let it test its Starship rocket again until late November at the earliest. "Environmental regulations and mitigations serve a noble purpose, stemming from common-sense safeguards to enable progress while preventing undue impact to the environment," SpaceX said in a statement. "However, with the licensing process being drawn out for Flight 5, we find ourselves delayed for unreasonable and exasperating reasons."
ON OUR FEED
“Generative AI is 90% of the airwaves and 5% of the use cases.”
—Gartner AI research chief Erick Brethenoux thinks generative AI is massively overhyped, with most businesses standing to benefit more from older AI techniques like machine learning and knowledge graphs, or in some cases a mix of those and genAI. He told The Register that any AI technique that continues to require specialist hardware, as gen AI mostly does for training and deployment, will ultimately be “doomed.”
IN CASE YOU MISSED IT
Apple hopes its iPhone 16’s AI features will be enough to revive flagging sales, by Jeremy Kahn
Ireland’s special relationship with Apple and Google has created a multibillion-dollar windfall—and an existential crisis, by Ryan Hogg
Inside X, fears rise of forced relocation to Texas with new HQ, by Kali Hays
Amazon to expand U.K. Operations with £8 billion investment in new AWS data centers over five years, by AFP
Gen Z moves on from ‘Googling’—TikTok emerges as the new search engine, by Sasha Rogelberg
GameStop reports unexpected profit despite fourth-straight quarter of declining sales, by Bloomberg
BEFORE YOU GO
Lawmakers vs AI. Eight Democratic senators have called for an investigation into the possible anticompetitive effects of new AI tools that summarize news articles and search results. “While a traditional search result or news feed links may lead users to the publisher’s website, an AI-generated summary keeps the users on the original search platform, where that platform alone can profit from the user’s attention through advertising and data collection,” they wrote to the Justice Department and Federal Trade Commission, according to TechCrunch.
