Lawmaker favors ‘issue by issue’ approach to AI regulation
Rep. Jay Obernolte (seated right) speaks with a colleague during a recent hearing on Capitol Hill. The California Republican urged Congress to move on artificial intelligence laws, and preempt states "issue by issue." Tom Williams via Getty Images
By
Chris Teale
Congress should not just pass sweeping federal standards and leave states to regulate the rest, Rep. Jay Obernolte, a California Republican and a key player on regulating the tech, told Route Fifty.
Lawmakers in Virginia recently passed a sweeping bill to regulate artificial intelligence, creating requirements for the development and deployment of high-risk systems, in a bill with echoes of the legislation vetoed last year in California.
It’s unclear whether that Virginia bill will meet the same veto fate at the hands of Gov. Glenn Youngkin. But the effort again highlights state governments’ desire to regulate AI in the absence of federal inaction on the subject and is kicking off what could be another busy year for the technology.
Amid that state-level activity, Congress is seemingly preparing itself for national regulations, which could preempt much of these new laws. The House Task Force on Artificial Intelligence issued a lengthy report on how lawmakers should approach the technology late last year laying out various options for preemption, including setting federal minimum and maximum standards.
That task force’s co-chair, Rep. Jay Obernolte, said he favors an “issue by issue” approach to preemption where those issues have been delegated to states, rather than setting broad federal standards. In an interview with Route Fifty, the California Republican said broad federal minimum or maximum standards could create a “patchwork” of state laws.
Given that AI is related to interstate commerce, and the U.S. Constitution explicitly gives Congress the power to regulate that, national lawmakers must step in, he said.
“I think it's a much better policy, instead of talking about floors and ceilings, to talk about issue by issue,” Obernolte said on the sidelines of the FedGIS Conference, hosted by leading geographic information system and location intelligence company Esri in Washington, D.C. “[For example], insurance rate regulation, and the use of AI to set insurance rates. Whose responsibility is it to oversee? That particular issue, is that a state issue? Is that a federal issue? If you can carve that out issue by issue, I think you create a road map that shows us where we need to be sure.”
An issue-based approach could prove complex, given the sheer number of areas that states and the federal government regulate, and that could be impacted by AI. While it may be tempting for Congress to just pass sweeping federal standards and then let the states get on with the rest, Obernolte said focusing on specific issues is doable.
“I think you do the best you can,” he said. “You carve out some issues, and you do that in a piece of legislation, and then you discover, ‘OK, well, there's some more issues we should have carved out.’ So that's another piece of legislation. Or you discover, ‘well, this particular issue we carved out, it's a little more complicated; we need to parse that down.’ And so you have a bill that modifies it.”
Obernolte has consistently called for Congress to act on national AI regulations, and said state leaders are only passing their own laws in the absence of federal action. He pointed to Colorado Gov. Jared Polis, who signed a sweeping state-level bill last year but in his signing statement said it would need edits. He also said Congress must step in “with a needed cohesive federal approach.”
Similarly, Obernolte said California State Sen. Scott Wiener, a former colleague of his in the California State Legislature, told him that he only felt “obligated” to propose AI regulations as Congress has so far done little on the technology.
“If you look at the states that are passing these laws, in many cases, they're very aware that this is something that it would be more appropriate for Congress to do,” Obernolte said. “But unfortunately, there isn't a lot of optimism that Congress is capable of acting on this piece of regulation.”
It remains to be seen what will happen on national AI regulations. In the meantime, California is already in the process of exploring a new bill on the technology. So far, it is just a preliminary measure designed to declare “the California Legislature's intent to develop future legislation.” This declarative bill shows a “commitment” to regulating areas like “technology governance, ethical considerations, potential risks, and regulatory frameworks.”
State Sen. Jerry McNerney, a Democrat who previously served five terms in the U.S. House before retiring in 2022, is behind that piece of legislation. A McNerney spokesperson said in an email that work is ongoing on the bill’s language, and it probably will be ready “in a few weeks.” Obernolte said Congress needs to get moving on its own efforts.
“I think it's appropriate to take an incremental approach, because AI is such a fast moving and rapidly changing field that the things that are appropriate today might not be appropriate tomorrow, but certainly it's true that we need to get started,” he said. “We're never going to finish if we don't start.”