California Democrat Rep. Ami Bera, M.D., said Congress needs to set up guardrails for healthcare AI, possibly following the legislative models set out by his home state.
Bera took the stage at the Washington Post Live’s Future of AI event on Tuesday morning in Washington, D.C. Bera participated in the bipartisan AI Task Force created by House Speaker Mike Johnson, R-La., and Democratic Leader Hakeem Jeffries, New York, last Congress.
Bera said the California congressional delegation has worked with its home state legislature to promote the passage of California state AI laws that could operate as a federal framework, once Congress decides to take up legislation.
“If California is ready to go, Colorado is also doing some things. Let's work with those states to adopt a framework that could easily be adopted at the federal level,” he said.
California has passed a number of state AI laws, including one that requires disclosure to patients when AI is used to communicate in clinical information, prohibiting healthcare payers from using AI for care denial, and protecting sensitive health data under its comprehensive state data privacy law. The Colorado Artificial Intelligence Act, passed in May 2024, seeks to prevent algorithmic discrimination in high-stakes decisions.
Bera did not mention if the specific pieces of legislation passed by California and Colorado are models for use at the federal level, though the lawmaker seems keen to jump on the forward momentum generated by the two states.
Bipartisan members of Congress came close to passing a ten-year moratorium on the enforcement of state AI laws over the summer during the budget reconciliation process. The proposal did not make it into the final version of the bill signed by the President on July 4, and Bera said he was glad it did not pass.
The AI law moratorium sought to halt a nationwide patchwork of AI regulation, which would be difficult for the healthcare industry to comply with—especially startups. Bera agreed there should be a national framework for health AI regulation, rather than a state-by-state approach.
Daniel Yang, vice president of AI and emerging technologies at Kaiser Permanente, said at an industry conference last week that the health system was tracking over 500 pieces of state AI legislation.
The Democratic doctor is not looking to give the Food and Drug Administration additional authority to regulate AI, and he said on stage that bipartisan members of Congress are in agreement. “The FDA probably has most of the tools and authorities that it needs to approve some of these algorithms,” he said.
The FDA has approved nearly a thousand medical devices that use AI, but the agency does not have oversight of administrative technologies like AI scribes.
Bera admitted that the FDA lacks the ability to keep up with the constantly evolving nature of generative AI, even in the products that it regulates. The representative said the agency should not have to re-approve algorithms on a regular basis, but he also asserted that the task can’t be left to industry to self-regulate.
He said ongoing monitoring of AI tools needs to land somewhere between regulation by the FDA and self-regulation by industry. He said the issue requires “urgency because we see how fast AI is evolving.”
Bera and other members of Congress are calling for the creation of a select subcommittee on AI so the public can have a window into Congress’ evolving knowledge on the topic. Bera said he expects the push for the select subcommittee to be renewed after Congress deals with its September 30 government funding deadline.
“I think it really should be a subcommittee where the public can actually see those experts coming in, the folks that have been briefing us, the questions that we're asking, because right now, the more questions we ask, the more questions we have,” Bera said. “Congress does have a role, and again, at a minimum, putting up guardrails.” If Congressional leaders don’t create the subcommittee, Bera said he would push for another task force.
The topic of physician liability for AI was raised, and the doctor-turned-Congressman said he believes physicians should be responsible for the recommendations they give to patients, even if a faulty recommendation was promoted by an AI tool. He also raised the issue of potential liability a physician may have for not using an AI tool that could have spotted human error, saying that the issues would likely be resolved in case law.
When asked about the Trump administration’s views on AI, promoted by the Centers for Medicare and Medicaid Services’ Administrator Mehmet Oz, M.D., for assistance with prior authorization, Bera said he largely agrees with the administration’s initiatives, calling them nonpartisan.
In a separate line of questioning, Bera condemned Health and Human Services Secretary Robert F. Kennedy Jr.’s actions on vaccines and undermining public health. He signed a letter last week that called for Kennedy’s resignation, the Washington Post moderator said.