Artificial intelligence used for political purposes would require a disclaimer under a bill proposed in the North Dakota Legislature.

The House Government and Veterans Affairs Committee held a hearing Friday on House Bill 1167 that would require a “prominent disclaimer” on any political communication or political advertisement created wholly or in part by artificial intelligence tools.

The disclaimer must read: “This content generated by artificial intelligence.”

Bill sponsor Rep. Jonathan Warrey, R-Casselton, said he expects AI use in political communications to become a much bigger issue in the future.

“There are other states taking action on this to provide some protections in place, and I think the overall theme is very sound,” Warrey said. 

He added the bill was crafted through consultation with the Secretary of State’s Office and creates a new provision under the Corrupt Practices section of North Dakota Century Code, making any violation punishable by a Class A misdemeanor.

Deputy Secretary of State Sandy McMerty testified in support of the bill. She said AI can be used to help create graphics, write a newsletter and other communication. 

However, if AI is used in political communication, she said the public should be informed. McMerty likened the new policy to statements politicians are already required to attach to their political ads that say who paid for the ad.

Terry Effertz, executive director of advocacy group TechND, testified against the bill, telling lawmakers the proposal is too broad.

“The bill, to be honest, is a reaction to hypothetical concerns, rather than a solution to documented harm,” Effertz said. “AI is evolving and hasty legislation in this area could inhibit legitimate uses while failing to address the actual risks.”

Because AI has become widely embedded in digital content and software, it could lead to “disclosure overload,” she said.

“Really what we need to focus on is the fact that deepfakes are the real threat here,” Effertz said.

A separate proposal, House Bill 1320, would outlaw the fraudulent use of deepfake videos and images in North Dakota. 

Deepfakes, or videos, images or recordings manipulated by generative AI, have caused concerns about spreading misinformation. The bill would make the creation, possession and release of deepfake videos and images, without the consent of the person featured, a Class A misdemeanor. 

Rep. Josh Christy, R-Fargo, the prime sponsor of the bill, said deepfakes are a threat to North Dakotans because it’s become more difficult to determine what is real and what is fake. 

He said the intent of his bill is to prevent someone from using someone else’s likeness without permission. The bill is not limited to deepfakes used for political purposes.

“If I’m able to take a video of you, upload it to a service, and then be able to represent you in a way that you don’t want, it’s not a good thing,” Christy said.

A public hearing on the bill is at 11 a.m. Monday in Room 327B at the Capitol. Christy said he plans to play a video of himself reading a portion of the Gettysburg Address in Russian, German and from a female avatar that he programmed though software.

He said he doesn’t want to cross any lines with satire or First Amendment concerns and hopes to get some feedback during the hearing.

“I don’t know where that line is,” he said. “Hopefully the Attorney General’s Office or others will come out for testimony on this and help clarify any amendments.”

The committee did not take immediate action on the bill related to disclosures of AI in political ads. Written testimony on the bill addressing deepfakes can be submitted online until 8 a.m. Monday.