State leaders raise concerns over AI-generated public comments in California

SACRAMENTO, Calif. — California lawmakers are debating how to regulate artificial intelligence, with growing concern over its potential to influence government decisions — particularly in public comment processes.

The bipartisan push centers on maintaining a clear distinction between humans and automated tools, arguing that the right to weigh in on government decisions belongs to actual people.

Public comment is “one of the most fundamental rights for the average citizen to weigh in on government decisions,” said state Sen. Christopher Cabaldon, a Democrat from Yolo who authored the legislation.

“It’s for people, not for robots or AI agents,” Cabaldon said.

California transparency laws, including the Public Records Act and Brown Act, are designed to ensure public access to government meetings, discussions and records, with some exceptions.

Lawmakers say they’re concerned those protections could be undermined if artificial intelligence tools are used to flood agencies with comments that appear to come from real people.

“That the public comment process doesn’t get hijacked by AI tools. That we really focus on real people,” said state Sen. Ben Allen, a Democrat from Pacific Palisades who is a co-author of the bill.

Reports suggest those concerns are already playing out in California, as AI platforms are used to mass-produce emails or comment letters intended to influence policy decisions.

Lawmakers argue that could overwhelm government agencies and drown out input from actual residents.

“There is this flood of popular support or opposition on a particular measure that’s not actually reflective of real people’s opinions,” Allen said.

Environmental groups say they have seen the impact firsthand.

“The Sierra Club has historically turned our membership out to provide comments and testify on state legislation, on agency rules and regulations,” said Dylan Plummer with the Sierra Club, the nation’s largest grassroots environmental activist organization.

The organization said its support for a measure to phase out gas-powered appliances in Southern California was overshadowed by more than 20,000 opposing comments generated by Civiclick.

But Civiclick strongly disputes that characterization. Its founder tells ABC10 though the company offers AI tools to clients, AI was never used in the specific scenario and that that all opposition letters came from real people, not bots.

Cabaldon has introduced Senate Bill 1159, a bipartisan proposal aimed at preventing the use of AI or related automated tools in the public engagement process.

The bill passed the Senate Judiciary Committee unanimously, though lawmakers raised concerns about how it would be enforced, and how the use of AI would be detected.

Cabaldon acknowledged that those are needed discussions as the bill advances.

“If I’d read this bill back when I was on the Sacramento County Board of Supervisors, I would’ve wondered what you were smoking, but my point is, that’s how things have progressed,” said state Sen. Roger Niello, a Republican from Fair Oaks, earning chuckles from fellow committee members. “I get what you’re trying to do, and I support that, but I think it’s going to be difficult.”

Some lawmakers also questioned whether local governments could face liability if AI detection systems fail and inadvertently exclude real people from participating.

“Is there any fear we’re actually creating a liability for local governments if their detection fails, and they omit an opportunity for a human?” said state Sen. Suzette Martinez Valladares, a Republican from Santa Clarita.

Cabaldon acknowledged the concern.

“I don’t know if I’d go that far but yes, to the general point, yes,” he said.

“That is one of the central questions. One is the detection side, but the complementary question is if you get it right 99% of the time, what happens to that 1%?” Cabaldon added. “is there any recourse to the human being who was overlooked?”

The bill now heads to the Senate Committee on Privacy, Digital Technologies and Consumer Protection.