The National Association of Insurance Commissioners is struggling to reach a consensus on regulating insurers' use of AI.
Speaking during S&P Global Market Intelligence's Insurance Investments Symposium on Oct. 9, Wisconsin Insurance Commissioner Nathan Houdek said that the Big Data and Artificial Intelligence Working Group of the National Association of Insurance Commissioners (NAIC) is looking at whether to move forward with "some type of model law or regulation." Houdek noted that the goal of new regulation from the NAIC is to have each of the states adopt it across the country to have as much uniformity as possible.
"That's where the discussion is right now: Is there overwhelming support to move forward with developing a model law?" Houdek said. "The jury is still out; you see a split in the NAIC membership. Some states think we need to be doing more, and then other states have a very different opinion."
The NAIC adopted its AI model bulletin — Use of Artificial Intelligence Systems by Insurers — in December 2023, and as of Aug. 5, there were 24 NAIC jurisdictions that adopted the bulletin. So far, Colorado, New York, California and Texas have their own state-issued guidance for insurers.
The NAIC's AI work
The NAIC AI model bulletin lays out several expectations for insurers, particularly that the industry's use of AI systems must adhere to legal and regulatory standards, as well as unfair trade practice laws.
"Because, in the absence of proper controls, AI has the potential to increase the risk of inaccurate, arbitrary, capricious, or unfairly discriminatory outcomes for consumers, it is important that insurers adopt and implement controls specifically related to their use of AI that are designed to mitigate the risk of Adverse Consumer Outcomes," the bulletin reads.
Insurers in states that have adopted the bulletin are also supposed to "develop, implement, and maintain a written program" — called an AIS Program — to address "responsible use of AI Systems that make, or support decisions related to regulated insurance practices."
The NAIC is also working to develop an AI systems evaluation tool to help regulators identify and assess risks stemming from insurers' use of AI. The draft of the tool depicts several questionnaires for insurers to respond to regarding AI utilization. The 60-day comment period for the tool ended Sept. 5.
The divide
During a Sept. 29 interim meeting of the NAIC's Big Data and Artificial Intelligence Working Group, regulators discussed what, if anything, should be done in addition to the AI model bulletin. It was apparent that commissioners fell on different sides of the issue.
Colton Schulz, chief examiner and company licensing and examinations division director at the North Dakota Insurance Department, said during the meeting that his state is working on its own version of the NAIC AI bulletin.
"We just have to get it in our language," Schulz said, adding that he would be interested in seeing an analysis on what an AI model law could provide that is not already covered by the bulletin.
Mary Block, deputy insurance commissioner with the Vermont Department of Financial Regulation, said Vermont has been "tossing around the idea of client disclosures, particularly around the use of data and what kind of data is being used." Block noted that Vermont has a "generic disclosure statute" that would enable its insurance commissioner to build regulation related to disclosures.
"We've been trying to decide whether to use it or not," Block said. "Obviously, it would be a lot easier if there was a model ... that we could borrow from, but that is definitely on the table for us."
During the same meeting, Iowa Insurance Commissioner Doug Ommen cautioned the group on moving too quickly.
"Certainly I think there's some interest in talking about a disclosure law, but frankly, my view is that moving forward right now might be premature," Ommen said. "Frankly what work we've done with the bulletin is really primarily grounded in the principles of governance and that along with this evaluation tool is really putting us as regulators in a good place to look at our own industries and better understand what it is that they're doing."
Colorado Insurance Commissioner Michael Conway countered Ommen's view, saying that it is "vital" that the group makes progress on AI regulation.
"It's vitally important for us to be moving forward with some action as it relates to model laws or model regulations in the AI space," Conway said. "If we don't, we will see other regulatory bodies, within our states and at the federal level, step into that space. So I think it's vital for us to fill that void."
Conway added that the movement to standardize disclosures makes for a "logical first step" but noted that he thinks the NAIC is "going to have to go further eventually."
AI regulation of the insurance industry is already on the federal government's radar, as evidenced by a late-July hearing of a subcommittee of the US Senate Committee on Banking, Housing and Urban Affairs.
During the hearing, senators discussed how insurers are using AI and the role that Congress should play in fostering innovation while protecting consumers with potential guardrails on AI use. Senators from both sides of the aisle agreed that there needs to be a regulatory framework of some sort to address AI issues.
Sen. Mike Rounds (R-SD), chairman of the Subcommittee on Securities, Insurance and Investment, said there needs to be additional clarity on how AI is used by companies, "whether by congressional action or through the work of standard-setting bodies."
In a recent interview with S&P Global Market Intelligence, South Carolina Insurance Director Michael Wise summed up the power of AI.
"Some people believe it's the best thing since sliced bread, other people feel like it's going to be this terrible thing," Wise said. "I would say it's a tool like any other. It depends on how it's used."
