COLUMBUS, Ohio — As artificial intelligence services become increasingly ubiquitous, several states are introducing bills to regulate certain uses of AI.
In Ohio, one AI-specific regulatory proposal, Senate Bill 163, has accrued some support from across the aisle and strong advocacy from Ohio Attorney General Dave Yost, the state’s top law enforcement official.
The legislation, introduced by Sen. Bill Blessing, R-Colerain Twp., and Sen. Terry Johnson, R-McDermott, focuses on AI’s ability to generate so-called deepfakes, which are fabricated videos, images or audio recordings that use someone’s likeness, often without the consent of that person.
S.B. 163 would do three main things.
First, it would require all AI-generated images to come with a watermark signaling the image is fabricated, and it makes it a civil violation to remove that watermark for the purpose of concealing where the image came from.
Second, the bill would make it a third-degree felony for a person to make or transmit any artificially generated depiction of a minor; and a fourth-degree felony to possess any “simulated obscene material” of anyone.
Third, the bill would expand Ohio’s definition of identity fraud, making it illegal to use a “replica of a person’s persona” (a person’s voice, photograph, image, likeness, or distinctive appearance) with the intent to defraud, pressure someone into a financial decision, damage a person’s reputation, depict someone in a sexually suggestive way, or “for the purposes of violating child enticement or child obscenity laws.”
The bill sits in the Senate Judiciary Committee and has had three hearings.
“I think there are a number of people that see the problems with creating something that looks exactly like somebody else in a compromising position, (or) using somebody’s voice to commit identity fraud,“ Blessing said in an interview. ”You know, there are all sorts of things that this can be used for. I think the General Assembly recognizes we’ve got to work on this.”
The state has two primary options on how to regulate actions like this: it could place restrictions on a consumer’s legal uses of a product, or it could place restrictions on what type of functions a producer’s product can actually produce.
Blessing said he wants to do both.
“It’s one thing to go after individuals. If somebody’s using AI to create child pornography, obviously, go after them — I don’t know that the tech industry would have any issues with that either,“ Blessing said. ”The problem is that there are only a handful of big platforms to disseminate stuff like that at a large scale; there’s only so many software platforms you can use to generate that sort of stuff, so I want to stay focused on going after the root.”
Blessing described this as an uphill battle, made abundantly clear with the tech industry’s recent failed attempt to pass a federal law that would have blocked states from passing new AI regulations for the next 10 years.
“Broadly speaking, the tech industry is very powerful for obvious reasons,” Blessing said. “Their economic might leads to massive political might, and they simply do not want to be regulated in any meaningful fashion.”
S.B. 163 has so far seen measured opposition from tech advocates.
A representative of TechNet, a lobbying firm for tech giants like Apple, Amazon, Meta, OpenAI and more, told the Senate Judiciary Committee that its members commend S.B. 163’s goals of “protecting children, preventing fraud, and promoting transparency in AI-generated media.”
Still, TechNet asked for various changes in the bill, including language clearly stating that “liability should be squarely on the bad actor: the person creating, viewing, promoting, or distributing” nefarious content.
The bill got a more full-throated endorsement from Yost, who bemoaned AI’s lack of transparency and told lawmakers that he viewed S.B. 163 as a measure that would protect individual liberties, Ohioans’ faith in government and provide safeguards for children in Ohio.
“Protecting the unprotected is key here. We absolutely must do everything we can to ensure that AI tools are not used for evil, and that our children can be children without the fear of AI manipulated photos appearing online,” Yost said. “These are the type of things that keep me up at night.”
