Health insurers have spent years using artificial intelligence to process claims after care already happened. That focus is moving upstream into decisions about which patients get flagged, which treatments get pre-approved and which care pathways get recommended before a bill is ever generated.
Sun Life disclosed in its second-quarter 2026 earnings release that it is now using AI-powered clinical navigation from Medzown, a precision medicine management company, to identify members diagnosed with cancer and other complex, costly conditions and connect them to an appropriate clinical trial “before high-cost claims escalate.”
The numbers explain why. Cancer costs an average of $252,000 per individual per year, and orthopedic and musculoskeletal conditions cost an average of $120,000, according to Sun Life’s own 2026 high-cost claims report. Sun Life’s broader suite of AI and digital health tools, combined with other cost containment efforts, saved the company and its self-insured employer clients more than $68 million in 2025, the company said in its earnings report.
“Employers who self-fund their health plans face immense financial risk when members develop complex, costly conditions,” Jennifer Collier, president of Health and Risk Solutions at Sun Life U.S., said when the Medzown partnership was announced. “Medzown’s clinical expertise and individualized support is uniquely suited to address both the human and financial sides of managing major health conditions.”
Aetna Is Automating Both Ends of the Same Process
Sun Life is not alone in pushing artificial intelligence earlier into the patient journey. Aetna, a CVS Health company, has spent the past year building AI directly into how members navigate care and how the insurer processes what care costs. Its Care Paths tool uses AI to give members personalized recommendations for health and wellness programs tailored to their specific conditions, and the company is embedding in a conversational AI assistant into the tool to guide members through those choices, according to CVS Health.
On the claims side, Aetna launched a second-generation version of its Claims Assist Manager in May, an AI-powered platform that reduces processing time by more than 20% for complex claims requiring manual review, CVS Health said. The launch is part of a broader $20 billion, multi-year investment in digital tools.
Together, the two tools illustrate the direction insurers are moving: AI that shapes what a member does before treatment, paired with AI that adjudicates what the insurer pays after it. That same infrastructure now touches both ends of a claim’s lifecycle, not just the the back-office processing AI in insurance was originally built for.
States Are Drawing a Line at the Coverage Decision Itself
That expansion has run directly into a wave of new state regulation aimed specifically at how far insurers can let AI go before a human has to step back in. Seven states passed new laws in 2026 addressing AI’s role in healthcare, with prior authorization the primary focus, according to a legislative review by Becker’s Payer Issues.
Alabama’s SB 63, effective Oct. 1, prohibits insurers from using AI as the sole basis for a coverage denial and requires disclosure whenever AI is used in the review process. Colorado’s HB 1139 requires that AI-assisted utilization review decisions be based on a patient’s individual clinical history rather than group data, and mandates that a licensed clinician personally review any medical-necessity denial before it takes effect. Washington’s SB 5395, effective June 11, 2026, goes further still. Only a licensed physician or health professional may deny a request on medical necessity grounds, AI cannot be the sole means used to deny, delay or modify care, and insurers must report to the state’s insurance commissioner what share of their prior authorization denials involved AI.
The pattern across nearly all of these laws is the same: insurers can use AI to flag, recommend and pre-screen, but the moment a decision actually denies or delays a patient’s care, a human must be the one who signed off on it. Whether that boundary holds as insurers push AI further upstream, into decisions that shape a patient’s treatment options before a formal denial is ever issued is yet to be seen.
