Health insurers are fighting a sharp rise in AI-generated fraud-fake medical records, synthetic identities, and bot-driven call campaigns-while liability coverage fails to keep up, exposing companies to what the market calls "silent AI" risk. The trend is pushing up claims screening costs and forcing insurers to rethink their fraud defences at a time when healthcare fraud already costs the industry tens of billions each year.
AI gives fraudsters a low-cost toolkit
Large language models can now produce convincing medical records for treatments that never occurred. AI bots can flood an insurer's call centre with thousands of calls without human involvement. Kurt Spear, vice president of financial investigation and provider review at Highmark, said the industry knew AI would be used by fraudsters and "now we're starting to see that." Highmark is deploying tools to detect AI-driven fraud and new technology to flag anomalies in medical imaging down to the pixel level, according to TribLive.
Call-centre threats are scaling fast. Pindrop's technology, used by major health insurers, tracked clients receiving 15,000 bot calls in just a few months, according to Jason Barr, Pindrop's vice president of healthcare. Yet the risk is unevenly distributed-UPMC and Pennsylvania's Medicaid agency told TribLive they have not yet seen much AI-driven fraud. The U.S. Department of Justice's 2025 national healthcare fraud takedown charged 324 people connected to over $14.6 billion in planned losses, a figure that generative AI is only likely to push higher.
Liability cover fails to price in AI exposures
Gallagher reports that many policy wordings do not directly include or exclude AI-related losses, creating a "silent AI" risk similar to the early days of cyber insurance. Research from Datacom shows that 87% of New Zealand organisations now use some form of AI, up from 66% in 2024 and 48% in 2023. About 20% of insurance professionals in Gallagher's latest adoption survey said insureds faced economic losses or made claims tied to AI risks over the past year. Of those, just over half were fully covered, 44% received partial coverage, and 3% got nothing.
Paige Cheasley, who leads Gallagher's Canada national technology practice, said, "it can be tricky to attribute losses directly to AI." That ambiguity leaves gaps across multiple lines-cyber, professional indemnity, product liability, and directors and officers (D&O) insurance. John Farley of Gallagher added that more AI use could increase both the frequency and severity of claims. The "silent AI" risk highlighted by Gallagher points to the need for clarity in AI for Insurance policies, as ambiguous exclusions could leave companies undercovered.
Automation is no substitute for human judgment
Even as insurers deploy AI for fraud detection, the technology is not a standalone solution. Janette Hiscock, who heads Europe, Middle East and Africa for UnitedHealthcare Global, told a Health & Protection briefing: "AI is not infallible, it is hackable." Algirdas Dineika of WTW stressed that insurers still need staff who can work alongside agentic AI systems. Relying too heavily on automated screening can lead to more false positives, slower legitimate payouts, and new privacy or fairness risks.
Joanne Buckle, principal and consulting actuary at Milliman, warned that AI advances are outpacing what organisations can manage. The National Association of Insurance Commissioners found that 65% of health insurers are using or planning to use AI or machine learning for fraud detection, a trend that reflects the broader application of AI for Healthcare in claims integrity. Those tools are supporting human decisions rather than replacing them, the NAIC report said.
Why this matters for healthcare
Healthcare organisations face a dual challenge: AI makes fraudulent claims cheaper and more detailed, but it is also one of the few ways to spot those scams at speed. Insurers that pair image forensics, voice authentication, and claims analytics with clinical review and clear audit trails will be better positioned than those relying on screening alone. For healthcare professionals, the message is clear-fraud controls must now account for AI-generated evidence, and liability policies need explicit AI language to avoid coverage surprises.
Your membership also unlocks:
