Undisclosed AI calls worry

Doctors and vendors are starting to use artificial-intelligence phone agents to call insurers on prior authorization and billing work, and lawyers say the biggest unanswered question is whether the caller has to say it is a machine. (fcc.gov) The Federal Communications Commission said on Feb. 8, 2024 that calls using AI-generated voices count as calls using an “artificial or prerecorded voice” under the Telephone Consumer Protection Act, which means prior express consent rules apply. (fcc.gov) The Federal Trade Commission said on March 7, 2024 that the Telemarketing Sales Rule bars deceptive robocalls, including voice-cloning calls, and requires telemarketers to make certain disclosures and avoid misrepresentations during sales calls. (ftc.gov) In healthcare, the calls at issue are not usually cold sales pitches to patients. They are routine contacts with payers over prior authorization, coverage checks, and claims status, in a system the American Medical Association said consumes about 12 hours a week of physician and staff time per doctor. (ama-assn.org) That workload is colliding with a federal push to modernize prior authorization. The Centers for Medicare & Medicaid Services finalized a rule on Jan. 17, 2024 requiring Medicare Advantage, Medicaid, Children’s Health Insurance Program, and certain exchange plans to build electronic prior-authorization tools, with compliance dates generally beginning in 2026 and 2027. (cms.gov) The legal gap is that federal rules clearly treat an AI voice as an artificial voice, but they do not yet set a simple nationwide script for healthcare callers that says when an AI agent must identify itself at the start of a payer call. The FCC opened a broader AI robocall proceeding in 2023 and said in 2024 that AI-generated voice calls create new consumer-protection questions under existing law. (docs.fcc.gov) State insurance and consumer regulators are already pressing for more transparency around AI in coverage decisions. A November 14, 2024 report submitted to the National Association of Insurance Commissioners said it “must be clear” when AI is being used in utilization management and what role it plays in coverage determinations. (content.naic.org) Insurers and providers are both automating faster. A Health Affairs analysis published in 2025 said 37% of insurers reported using AI, or planning to within a year, for prior authorization, 44% for claims adjudication, and 56% for utilization-management activity. (healthaffairs.org) That leaves clinicians and software vendors with two layers of risk. One is regulatory enforcement if an undisclosed AI caller is treated as deceptive; the other is civil litigation, including the kind of class-action theories already being tested in other healthcare AI disputes over denials and recordings. (ftc.gov; healthcarefinancenews.com) For now, the safest reading is narrower than the technology. AI can place the call, but the law is moving toward more consent, more disclosure, and more documentation of who—or what—is speaking on the line. (fcc.gov; content.naic.org)