From meeting energy demands to mental health care, the state of Utah has been working to keep up with developments in artificial intelligence. But lawmakers know AI technology can have drawbacks, too.
Lawmakers want to reduce the risk that comes with AI’s generative abilities by expanding the definition of identity abuse to include using someone’s artificially recreated identity for commercial purposes without that person’s consent. That includes using their name, image or voice.
Republican Sen. Mike McKell, sponsor of SB271, told the Senate Judiciary, Law Enforcement, and Criminal Justice Committee on Feb. 19 that the legislation came from a Tennessee law. He said it’s meant to protect talent, artists and their work.
Utah researchers have kept an eye on the issue. Brandon Amacher, director of the Emerging Tech Policy Laboratory at Utah Valley University, said the ability to forge an individual’s image without their permission is concerning.
“There have already been several, several instances of really high profile people saying a bunch of people have been sending me these ads saying that I'm endorsing this supplement, or I'm endorsing this wellness coaching or something like that [and] I've never even heard of this,” he said.
The laboratory’s team specifically looked into how people are deceived by AI-generated content. They conducted a study that presented people with deepfake and authentic audio and video in an environment like they would see on social media. Of the participants, 50% thought the deepfake content they encountered was real. In fact, the deepfake video outperformed the real video, as participants found it more knowledgeable, trustworthy and persuasive.
“Some of the deepfakes that we use in our study my students created in one weekend with free software that they found online,” Amacher said. “It was literally just open source out there on the internet. They found it. They got it done.”
When this kind of content is used to sell products or ideas, multiple concerns can arise. On top of simple false advertising, Amacher said it can be used for extortion, embarrassment and spreading misinformation.
Steven Lund is the Chief Commercial Officer for SureMark Digital, a Utah-based start-up working to protect people’s identities from deep fakes. In his experience working with AI, he’s learned it can do a lot of good, but he still thinks there needs to be guardrails.
“There's a role for the government to play in this, primarily more around preventing it from happening,” Lund said. “You know, if somebody gets caught doing this, there's a significant repercussion for impacting a company or an individual's name, image or likeness.”
AI is rapidly advancing and that complicates any proposed restrictions. A problem or solution identified today can have something new come up tomorrow. So, Lund said there’s a need for continuous dialogue between lawmakers and industry leaders.
Amacher said McKell’s bill is a good first step, but a crime perpetrated through a digital medium needs a broader, national approach.
“Sometimes the victim might be in Utah, but how often are you going to run to a case where both the victim and the victimizer are in the same state and fall under that same legal framework,” Amacher asked. “How do you enforce this Utah law on them? That can be really tricky.”
Amacher knows artificially produced identity abuse is a novel issue. That’s why he said lawmakers need to be prepared to improve regulations as technologies change.
SB271 passed the Senate with a unanimous vote Feb. 27. It now heads to the House for further debate.
Elle Crossley is an intern with Amplify Utah and KUER covering the Utah State Legislature and other local news.
