![](https://media.npr.org/assets/img/2023/05/05/screen-shot-2023-05-05-at-1.07.14-pm_custom-2c8c68b26e712b479debf5b23445895ace0dd39c-s1000.jpg)
In 2016, Elon Musk went on stage at a tech conference outside Los Angeles and made a bold statement about the self-driving capability of Teslas.
"A Model S and Model X at this point can drive autonomously with greater safety than a person. Right now," the CEO told the Code Conference audience during a Q&A session following his interview with tech journalists Walt Mossberg and Kara Swisher.
Video of the session has been up on YouTube for nearly seven years. But it recently came back into the spotlight as part of a lawsuit brought by the family of a man who died when his Tesla crashed while using the self-driving feature. The family's lawyers cited that 2016 claim, along with others Musk has made about Tesla's self-driving software.
But the carmaker's lawyers pushed back.
Musk, "like many public figures, is the subject of many 'deepfake' videos and audio recordings that purport to show him saying and doing things he never actually said or did," they wrote in a court filing, going on to describe several fake videos of the billionaire.
Thanks to advances in artificial intelligence, it's easier than ever to create images and video of things that don't exist, or events that never happened. That's spurring warnings about digital fakery being used to spread propaganda and disinformation, impersonate celebrities and politicians, manipulate elections and scam people.
But the unleashing of powerful generative AI to the public is also raising concerns about another phenomenon: that as the technology becomes more prevalent, it will become easier to claim that anything is fake.
"That's exactly what we were concerned about: that when we entered this age of deepfakes, anybody can deny reality," said Hany Farid, a digital forensics expert and professor at the University of California, Berkeley. "That is the classic liar's dividend."
The liar's dividend is a term coined by law professors Bobby Chesney and Danielle Citron in a 2018 paper laying out the challenges deepfakes present to privacy, democracy, and national security.
The idea is, as people become more aware of how easy it is to fake audio and video, bad actors can weaponize that skepticism.
"Put simply: a skeptical public will be primed to doubt the authenticity of real audio and video evidence," Chesney and Citron wrote.
So far, courts aren't buying claims of deepfaked evidence
In Musk's case, the judge did not buy his lawyers' claims.
"What Tesla is contending is deeply troubling to the Court," Judge Evette Pennypacker wrote in a ruling ordering Musk to testify under oath.
"Their position is that because Mr. Musk is famous and might be more of a target for deep fakes, his public statements are immune," she wrote. "In other words, Mr. Musk, and others in his position, can simply say whatever they like in the public domain, then hide behind the potential for their recorded statements being a deep fake to avoid taking ownership of what they did actually say and do. The Court is unwilling to set such a precedent by condoning Tesla's approach here."
The Tesla lawsuit is not the first time deepfakes have been invoked in an attempt to rebut evidence — and likely won't be the last.
Two of the defendants on trial for the January 6th riot tried to claim video showing them at the Capitol could have been created or manipulated by artificial intelligence. Lawyers for one cited a 2017 deepfake of former President Barack Obama, created by researchers, as reason for the court to doubt footage of the riot that had been taken from YouTube.
But ultimately, both men were found guilty.
"So far, those seem to have been Hail Mary passes that have not succeeded," said Riana Pfefferkorn, a research scholar at the Stanford Internet Observatory who warned about the looming impact of deepfakes in the courtroom in a 2020 law review article.
While the threat that deepfakes could be offered as evidence is real, she said, "the courts have built up hundreds of years of resilience against efforts to introduce fake or tampered with evidence, going all the way back to faking somebody's signature on a handwritten document and going up through typed or mimeographed documents, film, video, Photoshopped evidence, and digital photography and videos."
In fact, Pfefferkorn thinks as AI technology proliferates, it's more likely that courts will confront accusations of fakery against real evidence than attempts to introduce fake evidence.
In those cases, she said, ethics rules and lawyers' professional norms have a role to play. Lawyers are not supposed to put forth frivolous arguments, for example.
"I think attorneys' own sense of self-preservation hopefully will go some distance towards incentivising them to do a little due diligence on the front end," Pfefferkorn said.
Juries may demand more proof
But these standards may need to be updated to specifically address what Loyola Law School professor Rebecca Delfino calls "the deepfake defense."
"Right now, it's sort of like the wild, wild west," she said, where lawyers can say, "'Well, let's run this up the flagpole and see what we can do with it.'"
Even as courts adapt to these challenges, the reverberations from AI fakes will still be felt.
If accusations that evidence is deepfaked become more common, juries may come to expect even more proof that evidence is real, in what Pfefferkorn compares to the "CSI Effect".
"If lawyers start to get juries to demand all the bells and whistles to prove that a piece of evidence is not a fake...that is a way for lawyers and for their clients who are seeking to downplay or dismiss damning evidence against them to essentially run up the bills and make it more expensive, more time-consuming for the other side to get that damning piece of evidence admitted," she said.
That could shut out people who don't have the resources to hire experts.
And whether inside or outside the courtroom, denying that real events actually occurred has corrosive effects. Today, most people carry around devices that can record what's happening around us at a moment's notice, enabling eyewitness accounts of protests, crimes, and public events to be easily shared.
Farid worries about a world in which people no longer believe documented evidence of "police violence, human rights violations, a politician saying something inappropriate or illegal."
"Suddenly there's no more reality," he said. "And that is really worrisome because I don't know how we reason about the world."