SC Statehouse lawmakers struggling to regulate AI even as its use and growth is expanding daily - Post and Courier

COLUMBIA — In today’s world, if you make a business phone call, should you have a legal right to know if the voice on the other end is live or AI?

That’s just one of the scenarios South Carolina lawmakers are grappling with as the artificial intelligence revolution goes into hyperdrive.

And what if the voice is so good it fools you, or says something incorrect or dishonest?

“It's a lack of (courage) to stand up and to protect our citizens,” Brandon Guffey, a Republican state representative from Rock Hill who has prioritized regulating tech companies, said in a recent interview about the Legislature’s AI problems.

“We're too worried about losing an arms race,” he added.

Across South Carolina, the use of AI is growing yet nothing in current state law requires its users to disclose whether they are AI or not. And, to date, little artificial intelligence legislation has moved in the Statehouse, even as public concerns around the technology mount with residents.

Questions about its use in industries, from mental health care to insurance, remain unresolved. And South Carolina has yet to develop a concrete definition for what artificial intelligence is — something a Statehouse committee discussed this week.

What’s been happening?

South Carolina lawmakers have been exploring what role AI will play in the state’s future, but so far haven’t done much about it.

The progress so far shows:

- Of some 31 pieces of Statehouse legislation proposed around artificial intelligence this past legislative session, just two became law, per data from the South Carolina Artificial Intelligence Observatory, which tracks state legislation on AI. Of the two that became law, both were related to criminal justice concerns, like the computer-aided generation of child sexual abuse images or the non-consensual creation of obscene AI-generated imagery.

- Bills regulating other uses of AI — restricting its undisclosed use in therapy, protecting children from chatbots or limiting the use of AI in determining someone’s eligibility for health coverage — stalled out without much discussion, either from industry pushback or a fear of dog collaring innovations in emerging technology. The state currently lags behind most others in the U.S. regarding AI legislation, according to the Council of State Governments.

- Two Statehouse committees formed to define artificial intelligence policies have also yielded little. A newly formed Senate artificial intelligence committee held its inaugural meeting Sept. 16 and will limit the scope of its work only to the incorporation of AI into state government. Another committee formed by House leadership in late 2023 to regulate AI has met infrequently on the subject since its inclusion into the workload of the existing House Regulations Committee.

That’s not to say nothing is happening. While it hasn’t produced much AI legislation, House AI Committee Chairman Rep. Jeff Bradley, R-Hilton Head, has been exploring methods to include artificial intelligence technology in South Carolina’s K-12 curriculum alongside other conversations about how the state should approach AI in the future.

“We are actively working on multiple initiatives,” Bradley wrote in a text message to The Post and Courier.

Speaking to reporters after the Sept. 16 first meeting of the special Senate Committee on Artificial Intelligence, committee Chairman Tom Young, R-Aiken, said the group is meant primarily to focus on the role of AI in government, rather than in everyday life. The members plan to meet into the next year, with plans to produce a final report and recommendations for state policymakers by the summer of 2028 on how the technology fits within the bureaucracy.

But Young acknowledged there was an appetite for them to do more.

“We've gotten emails in the last 24 hours ahead of this committee that came from various citizens and businesses,” Young told reporters. “A lot of people didn't really understand the scope of what we were here to do.”

New urgency

Artificial intelligence usage in South Carolina still lags behind the rest of the country. Data published by software giants Microsoft show the Palmetto State bearing relatively minor shares of its workforce utilizing AI technologies compared to other states, with the brunt of usage seen in Richland, Charleston and Oconee counties, all of which are home to major research institutions.

But the topic remains top of mind for many.

Political candidates for state and federal office have made AI and the data centers that power them central to their campaign messaging. Courtney McClain, a young Democrat running against Republican incumbent William Timmons in the Greenville-based 4th Congressional District, recently held a joint online rally on the topic with a candidate from neighboring Georgia, while state Sen. Josh Kimbrell — a member of the Senate committee on AI — recently held a town hall in his district specifically on data centers.

Voters are thinking of it, too.

“Why our constituents care about this issue, why they are glad we're doing this, is that they want us to get involved before the toaster burns the house down,” Sen. Matt Leber, R-Johns Island, said during the Sept. 16 hearing.

Other researchers, like Bryant Walker Smith from the University of South Carolina School of Law, agree, noting during the Sept. 16 hearing that some groups have spent years merely trying to come up with a working definition for what artificial intelligence even is, with lingering questions around whether our laws would be “resilient and flexible enough” for the anticipated and unanticipated issues around AI.

“If you get stuck in the weeds, you're not going to accomplish much,” said Homayoun Valafar, a professor of computer science and biomedical engineering in the College of Engineering and Computing at the University of South Carolina. “If it was up to me, I would focus on a very specific application of AI where it is of critical importance to develop policy.”

Where to start

Generative AI — using the technology to create images or campaign advertisements, for example — might be easier to regulate for the sheer reason that the hypotheticals around other uses of the tech are endless.

Are AI agents liable for crimes they commit, the people who created them or the software that enabled the creation of the agent in the first place? Can someone sue an insurance company if their claim was delayed by AI? Asking any of those questions, policymakers say, opens a new Pandora’s box too risky for concrete legislation, potentially putting at-risk the technology’s growth or the industries relying on it.

“I think we do a decent job once we're able to find a problem,” said House Majority Leader Micah Caskey, R-West Columbia. “For example, AI-generated child pornography: we decide you can't do that. Let's ban that. … We can do that. But as you move up the the hierarchy of uses, then it gets a lot harder to regulate it.”

The reluctance to regulate AI has frustrated some members of the legislature like Guffey, who has become a leading voice for tech industry reform after his son killed himself after being extorted via a social media application a half-decade ago. A member of the AI committee, he said he had grown frustrated with the committee’s work because of the fear of overregulating the emerging tech — even as they’ve identified issues needing to be solved.

“I'm a firm believer in laws should be reactive as opposed to proactive,” said Guffey. “We can sit here and think of all these things that could happen, but it doesn't matter. AI is not going to follow the law anyway, as long as there’s no liability there or financial penalties. That's about as far as you can regulate them.”