Comment from Felder, Kristen

I urge the Federal Trade Commission to investigate the use of artificial intelligence in insurance claims handling. The concern is not simply that insurers use AI. The concern is that multiple AI systems are now embedded inside insurer-designed claims processes that can restrict what information is collected, what employees see, what authority they have, and how easily they can challenge an automated result. A single claim may pass through several systems: one interpreting the First Notice of Loss, another summarizing facts, another classifying severity, another routing the claim, another generating an estimate, another evaluating fraud or liability, and another suggesting coverage or closure. Each output can narrow the information and choices available to the next system or employee. By the time a person called an “adjuster” reviews the claim, the result may already have been framed by several automated conclusions and insurer-selected rules. The insurer may say AI merely “assisted” a human, but that is not meaningful oversight if the employee lacks the training, complete information, or authority necessary to disagree. Several warning signs justify deeper investigation: • National homeowners claims closed without payment increased from approximately 25.7% in 2004 to 42.1% in 2024. • New York private-passenger auto liability claims closed without payment increased from 33.6% in 2005 to 48.6% in 2025. • Texas Watch found that, among 1,246 disputed auto claims that reached Right to Appraisal, independent review increased repairable claims by an average of approximately $5,307, or 131%, and total-loss claims by approximately $3,889, or 26%. • In one AI-handled auto claim described in my attached statement, the insurer estimated $438.10 on a policy with a $500 deductible. A qualified repair facility later estimated the damage at $2,697.40. The loss was listed as "below deductible" and the file closed. • The California Department of Insurance reported at least one alleged violation in 114 of 220 State Farm wildfire claim files reviewed—51.8% of the sample—with 398 alleged violations involving issues such as inadequate investigations, low settlement offers, underpayments, delays, and denials. • Current Oklahoma lawsuits against State Farm and Allstate separately allege wrongful underpayment or denial of claims, restrictive internal standards, and limitations on adjuster authority. These facts do not prove that every AI-assisted claim is deceptive. They do show enough recurring concern to justify a coordinated federal investigation. This issue cannot be addressed adequately through state regulation alone. Modern claims operations are remote, consolidated, and multi-state. The same vendors, models, software platforms, scripts, and authority structures may affect claims nationwide, while each state regulator sees only part of the pattern. The FTC should investigate the complete claims architecture, not just one algorithm: AI inputs + AI outputs + insurer-specific programming + workflow design + employee training + adjuster authority + escalation requirements + performance incentives. The Commission should also work with state departments of insurance and attorneys general, which already possess claim files, market-conduct findings, corrective-payment data, internal procedures, and vendor information. I respectfully ask the FTC to establish a federal baseline for: • Accuracy measured against the complete consumer outcome • Disclosure of material AI involvement • Disclosure of insurer-specific rules or steering • Meaningful human review by trained and empowered employees • Complete audit trails • First-inspection quality metrics • Review of claims consumers did not challenge Insurance should work fairly before a third-party appraiser, repair professional, public adjuster, regulator, or attorney becomes involved.Consumers are entitled to accurate fair intial assesments of damage. Fair claim payment cannot depend on the squeaky wheel. My full statement is attached and provides the supporting data, examples and specific investigative recommendations.