WASHINGTON (CN) — When Jennifer DeStefano decided at the last minute to answer a call from an unknown number, she was thrust into a harrowing ordeal that would put her at the center of an ongoing national conversation about the dangers of artificial intelligence.
With her husband and older daughter Brianna out of town for a skiing race, DeStefano, of Scottsdale, Arizona, picked up the phone out of an abundance of caution, she told lawmakers on the Senate Judiciary Committee’s subcommittee on human rights and the law during a hearing Tuesday afternoon.
“At the final ring, I chose to answer it, as unknown calls can often be a hospital or a doctor,” DeStefano said. But it wasn’t a hospital. It was DeStefano’s daughter, Brianna — or at least a voice that sounded exactly like hers.
DeStefano explained to lawmakers that the voice she believed to be her daughter, sobbing over the phone, begged for help before a man on the line claimed to have kidnapped her. The would-be kidnapper at first demanded $1 million in ransom for Brianna’s safe return, but later brought his demands down to $50,000.
At first, there was no reason to believe the voice on the other end of the line was not her daughter’s, DeStefano said. “It wasn’t just her voice, it was her cries, it was her sobs.”
DeStefano started negotiating with the man she believed had kidnapped her daughter, who demanded that she appear in person with the money. “They required me to get into a van with a bag over my head, with $50,000 in cash, to be transported to my daughter,” she said. “If I didn’t have all the money, then we were both going to be dead.”
Before she could follow through on those demands, though, DeStefano heard from her husband, who was safe with Brianna. The voice she believed to be her daughter’s was instead a convincing clone, generated by artificial intelligence.
DeStefano’s ordeal comes as Congress is working to get out ahead of emerging AI technology, which some lawmakers have framed as an effort to learn from the regulatory missteps of the early social media era. The Senate’s judiciary panel has taken a leading role in this campaign, holding a series of hearings in recent weeks aimed at unpacking the various threats posed by unchecked artificial intelligence.
The committee last week heard from witnesses who sounded the alarm about the impacts of AI on copyright and patent law. Lawmakers in May also grilled Sam Altman, founder of OpenAI, who said he would be open to federal regulation of his company’s technology.
During Tuesday’s meeting, DeStefano and a panel of AI experts warned that, if left unregulated, the developing technology could erode some of the fundamental pillars of society.
“The newest wave of generative AI is poised to fundamentally transform our collective sense-making,” said Aleksander Madry, a computing professor at the Massachusetts Institute of Technology.
AI tools allow people to generate content that is both realistic and persuasive to a general audience, Madry reasoned, and the technology is already cheap to deploy and broadly accessible. These factors are already making spam and phishing operations easier to conduct. AI is also making these scams more effective, the professor added, personalizing content to individual targets.
Generative AI models could also wreak broader havoc on the informed public, spreading disinformation and generating fake content, the experts warned. One place this can be particularly damaging is in the context of democratic elections, said Alexandra Givens, CEO of the D.C.-based Center for Democracy & Technology.
“In past elections, operatives used robocalls and texts to spread deceptive information,” Givens said. “But now, bad actors can easily use AI to exponentially grow and personalize voter suppression or other targeting.”
Givens also pointed to AI-powered government surveillance tools such as facial recognition as another potential problem spot. In the U.S., she said, the consequences of poor regulation for such technologies is already being seen — a Georgia man in late 2022 was jailed for nearly a week after facial recognition software falsely identified him as another individual wanted for a series of thefts in Louisiana.
“There are other accounts of wrongful arrests, and these are likely just the tip of the iceberg,” Givens said.
There are policy options for mitigating the drawbacks of AI, Madry told lawmakers. Congress should require companies deploying artificial intelligence for consumer use to label their products as AI-powered, he suggested. Regulators should also mandate those firms implement mechanisms for reporting improper use of artificial intelligence.
Lawmakers can also help shed light on the legal issues surrounding AI-powered crimes, Givens added. “Courts are going to have to tackle how the laws apply to new fact patterns, and whether and when AI companies bear liability for the content their tools produce versus downstream users.”
Finally, a broader public understanding of AI is also important, Madry contended. “I think no matter what happens, the public needs to understand how to interact with AI systems, and to be on the lookout for when they are actually interacting with AI in the first place. We do not want to learn this the hard way.”
DeStefano, who did learn the hard way about the dangers of AI, implored Congress to take swift action.
“Congress has a larger looming task ahead,” she said. “How do we move forward as a community with this haunting reality that is plaguing us? If left uncontrolled, unregulated, and unprotected … it will rewrite our understanding and perception of what is — and what is not — truth.”
Follow @BenjaminSWeiss
Subscribe to Closing Arguments
Sign up for new weekly newsletter Closing Arguments to get the latest about ongoing trials, major litigation and hot cases and rulings in courthouses around the U.S. and the world.
Additional Reads
-
Haiti transitional council sworn in after months of violence April 25, 2024 -
