Georgia lawmakers have moved to put a legal backstop between algorithms and patients’ medical coverage. Senate Bill 444, sponsored by retired orthopedic surgeon-turned-state-senator Kay Kirkpatrick, passed the Georgia Senate unanimously and cleared full legislative approval on March 31, 2026, before heading to Governor Brian Kemp’s desk. The bill amends the state’s private review agent statute to declare that insurance coverage decisions cannot be based solely on AI systems or other automated software tools, with an effective date of January 1, 2027.
What the law actually requires
Under SB 444, a qualified human reviewer must examine the clinical information behind any denial before it can be finalized — AI can flag or triage a claim, but it cannot be the final word that keeps a patient from receiving care their doctor ordered. “There needs to be a human review of the information before they turn you down for your treatment that your doctor ordered,” Kirkpatrick said in describing the bill’s intent, framing it as a response to constituent complaints about opaque, fast-turnaround denials. Kirkpatrick’s background as a practicing surgeon gave the bill added credibility with colleagues in the legislature, who heard repeated accounts of denials issued within seconds of a claim’s submission — a turnaround time patient advocates argued was implausible for genuine clinical review by any human, automated or not.
Part of a broader wave of state action
Georgia is not acting alone. According to a Becker’s Payer Issues review, seven states enacted AI-related health insurance laws in 2026, including Illinois’s SB 3114 targeting AI-driven “claim downcoding” and Alabama’s SB 63, which regulates how insurers use AI in coverage authorization determinations. KFF’s tracking of federal and state consumer protections found that at least 25 states have issued guidance built on a 2023 model bulletin from the National Association of Insurance Commissioners, reflecting a fast-building patchwork of rules aimed squarely at algorithmic claims review. That patchwork means a national insurer could soon be operating under materially different AI compliance obligations in Georgia, Illinois, and Alabama than in states that have not yet legislated, complicating what was originally sold to insurers as a uniform, cost-saving automation strategy.
The scale of AI adoption that prompted the pushback
The urgency behind these bills is rooted in numbers: an NAIC survey of 93 large health insurers found 43% already use or plan to use AI for claim adjudication, and 26% use or plan to use it for benefit determinations. Meanwhile, UnitedHealth Group has drawn sustained scrutiny — congressional inquiries have cited reports that its AI-assisted denial rate has, in some circumstances, run as much as 16 times higher than typical, fueling lawmaker anxiety nationwide, not just in Georgia. Members of Congress have separately pressed health insurance executives, including at UnitedHealth, on whether their companies can commit to safe AI practices in claims handling, testimony that has kept the issue in the national spotlight well beyond state capitols.
Industry pushback and the insurer’s counterargument
Insurance industry groups have generally argued that AI tools speed up legitimate claims — UnitedHealth, for instance, says it is spending roughly $1.5 billion on AI in 2026 and processing prior authorizations through a digital tool with a 96% approval rate, framing automation as a way to get more patients faster “yes” answers rather than more denials. Critics counter that speed and volume are exactly the problem: a system optimized to clear claims quickly can just as easily be tuned to reject them quickly, with far less scrutiny than a human adjuster would apply, especially at scale across millions of claims.
What happens after January 2027
Once SB 444 takes effect, Georgia regulators will need to define what counts as adequate “human review” — a fast rubber-stamp of an AI recommendation could satisfy the letter of the law while doing little to change outcomes in practice, a concern patient advocates have raised about similar rules elsewhere. Other states are expected to introduce comparable bills in 2027 legislative sessions, and insurers operating nationally will likely have to build state-by-state compliance workflows rather than a single AI claims pipeline, adding cost and complexity to a technology that was originally sold as a way to cut costs. Whether Georgia’s law meaningfully reduces wrongful denials, or simply adds a procedural checkbox that insurers satisfy without changing underlying decision logic, will likely become clear only after the first wave of post-2027 denial data is analyzed by state regulators and patient advocacy groups.
