AI is a powerful tool. And while it’s promising in some areas, it can also be used to deepfake faces and voices in ways that can teeter on the edge of election interference and legality.
Hence, Sioux Falls Democratic Sen. Liz Larson is bringing her bill to begin establishing code for the burgeoning tool with SB 148.
“We don’t have any statutes right now that address anything related to artificial intelligence and or deepfakes," Larson said. "This is considered the baby step toward putting the definitions into code and also dealing with a very real and present problem that could face anyone in an election cycle.”
That problem being the spreading of misinformation during an election cycle.
“It could completely change the outcome of an election," Larson said. "I gave the example of robocalls in New Hampshire, nobody was really sure if it was pro-Democrat or Pro-Republican. A lot of times, these deepfake-type events, they’re kind of just designed to sew chaos.”
While the bill passed out of the Senate, some lawmakers voted against this bill on the grounds of the First Amendment. Larson says her bill is entirely First Amendment compliant.
“It has an exception for parody and satire, and it also is not a ban," Larson said. "It’s just a requirement to disclose something was generated by AI. With those two things in mind, we were able to not violate the First Amendment.”
Other troubling uses for AI and deepfake generated images, such as revenge porn, remain uncodified in South Dakota law.