Democratic and Republican state legislators across the US are following California's lead and introducing bills to stop health insurance companies from only using artificial intelligence to deny claims. It's another example of states stepping in to regulate AI as the Trump Administration rolls back regulations introduced during the Biden Administration. “One day we're going to be comfortable with AI deciding our fates,” said Arizona Rep. Justin Wilmeth, a Republican. “We’re not there yet.”
California's AI regulation is having a ripple effect as both Democratic and Republican state legislators across the US take aim at health insurance companies’ use of artificial intelligence to deny claims.Last year, California passed the "Physicians Make Decisions Act” which prohibits health insurance companies from solely using AI to deny health insurance claims. This year, at least 11 states are proposing similar prohibitions in another example of states stepping in to regulate AI as the Trump Administration rolls back regulations introduced during the Biden Administration (see here).
Lobbyists for the health insurance industry pushed back in public hearings this week, arguing the bills go too far or aren’t needed. But in a sign of their challenges ahead, they couldn’t stop an Arizona bill from moving forward out of committee this week.
“One day we're going to be comfortable with AI deciding our fates,” said Arizona Rep. Justin Wilmeth, a Republican. “We’re not there yet.”
California’s AI law ensures that any denial, delay or modification of health care is made by a licensed physician or other healthcare provider qualified to evaluate the case. The law also commits companies that use AI in their utilization review process to fair and equitable standards.
Now at least 17 bills that include provisions prohibiting health insurance companies from only using AI to deny a claim or prior authorization are pending in Arizona, Connecticut, Illinois, Maryland, Nebraska, New Jersey, New York, Nevada, Rhode Island, Texas and Washington state. Four separate bills have been introduced in Connecticut.
Several bills also include disclosure, reporting and audit requirements, such as LB77 introduced in Nebraska by Sen. Eliot Bostar, a Democrat. In addition to prohibiting the sole use of AI to deny claims, Bostar’s bill would require disclosures to healthcare providers, enrollees and on public websites if AI is used to review health insurance claims. The state would be allowed to audit automated review systems at any time.
A bill in Maryland (HB697), backed by a bi-partisan group of House legislators, would also require health insurers to submit quarterly reports to the Maryland Insurance Commissioner about their use of AI and how many times AI was used in claim denials.
Lobbyists from the health insurance industry pushed back at some of the first public hearings on the proposals this week. But an Arizona bill sponsored by Rep. Julie Willoughby, a Republican, advanced over their objections.
Willoughby’s HB 2175 would require healthcare providers to individually review each claim before an insurer denies a claim or a prior authorization. It prohibits the use of AI when a denial involves medical judgment.
But the proposal is unnecessary, said Chad Heinrich, a lobbyist for Arizonans for Affordable Health Coverage, who testified in opposition to the bill. There are already statutory requirements that a licensed professional review every claim, he said.
It’s also too broad and could sweep in technologies that aren’t AI and that health insurers need, he said. “Even having like an e-mail sort function could be considered AI,” Heinrich said.
AI has many beneficial uses in health care, said Christina Corieri, a lobbyist for the Arizona Medical Association who testified in support of the bill. But “it can’t exercise medical judgment,” she said. “It can’t understand the nuance of a patient’s condition.”
She cited a study that found an insurer had created a program that allowed doctors deny to claims without actually ever opening patients’ files.
“This company allowed the doctors to deny 300,000 claims over a couple of months at a speed of one every 1.2 seconds,” she said.
The House Commerce Committee approved new language clarifying that AI may not be used to deny a claim or a prior authorization “for medical necessity, experimental status or any other reason that involves the use of medical judgment.” Then the members voted unanimously to move the bill forward on Feb. 4.
Insurance industry lobbyists also testified against a bill in Washington state at a public hearing on Feb. 5 that includes similar provisions around AI.
Under HB 1566, introduced by Rep. Alicia Rule, a Democrat, health insurers couldn’t use AI alone to deny, delay or modify healthcare services. Their policies and procedures for using AI or other software tools would be audited by the state’s insurance commission, which would review an AI tool's performance, use and outcomes for accuracy and reliability.
“This bill would bring clarity and transparency, and ensure decisions are made by humans, so people can get the medical care they expect and deserve,” Rule said.
Adam Dittemore, a manager with Evergreen Health, a nonprofit healthcare provider in Washington state, testified that there’s been a more than 40-percent increase in denials of requests for prior authorization. Those denials lead to lengthy appeals, which can lead to significant delays in healthcare, he said.
“We believe that’s due to an increasing use of AI to auto-deny requests,” Dittemore said.
But Jennifer Ziegler, a lobbyist for the Association of Washington Healthcare Plans, said the bill is unnecessary and would discourage investments in AI that could help patients.
“We don’t want this bill to make decisions that take us backwards in the investment of tech that can be better for patients,” she said.
Please e-mail editors@mlex.com to contact the editorial staff regarding this story, or to submit the names of lawyers and advisers.