Core Regulatory Finding
Software vendors advertising “fully automated settlements” or “zero-touch claim resolution” are marketing functionalities that constitute statutory adjusting under state law. Because algorithms cannot hold insurance adjuster licenses, insurers deploying unreviewed, autonomous claim valuation or denial engines face substantial regulatory sanctions and voided releases.
The Pitch vs. The Statute
Over the last 24 months, a wave of early-stage insurtech startups has flooded the property and casualty market with marketing promises that claim handling can be fully automated. Promotional materials boast of “instant settlement offers generated in seconds,” “autonomous liability determination without human bottleneck,” and “touchless bodily injury workflows.”
While attractive to venture investors seeking software-style profit margins, these claims ignore the bedrock legal framework governing insurance claims in the United States. Under state insurance codes, insurance adjusting is not a software workflow—it is a heavily regulated licensed profession.
The Statutory Line: Clerical Automation vs. Discretionary Adjusting
State laws draw an unambiguous line between permissible clerical data extraction and statutory adjusting activities:
| Activity | Clerical Technology (Permissible) | Statutory Adjusting (Licensure Mandatory) |
|---|---|---|
| Document Processing | Optical character recognition (OCR), converting TIFFs to searchable PDFs, indexing medical bill dates. | Determining whether medical records establish a causal nexus between the collision and claimed injuries. |
| Medical Billing | Extracting line-item charge totals, identifying provider tax IDs, and organizing chronologies. | Deciding whether treatment was medically necessary or applying arbitrary non-evidentiary billing reductions. |
| Liability & Coverage | Transcribing police report summaries and witness contact numbers into fields. | Apportioning comparative fault percentages (e.g., 80/20) or determining applicability of policy exclusions. |
| Settlement Value | Displaying demand amounts and verified itemized totals alongside each other. | Calculating binding valuation ranges, formulating counter-offers, or issuing settlement checks. |
Under leading state statutes, such as California Insurance Code § 14021, New York Insurance Law § 2101(g), and Texas Insurance Code § 4101.001, anyone who investigates, negotiates, or settles claims on behalf of an insurer is an adjuster. When a software platform determines liability, decides what a claim is worth, or generates an automated compromise offer without independent human adjudication, the software is practicing insurance adjusting without a license.
The Legal Trap for Carriers: Vicarious Liability and Bad Faith
Insurers often assume that because they purchased software from a third-party vendor, the vendor absorbs the legal risk. State insurance departments and appellate courts reject this assumption completely.
“An insurer is responsible for the actions of its third-party service providers, including the use of AI Systems on behalf of the insurer, to the same extent as if the insurer had taken the action directly.”
— NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, Section 1
When an insurer relies on an autonomous AI system that makes an adjusting determination:
- Administrative Sanctions: The state Department of Insurance (DOI) can initiate market conduct proceedings against the insurer for utilizing unlicensed adjusters, levying fines and ordering mandatory remediation audits.
- Voidable Releases: Plaintiff personal injury attorneys who discover that a settlement offer was generated autonomously can move to set aside signed releases, alleging deceptive trade practices, lack of informed negotiation, and unconscionable adjusting practices.
- Catastrophic Deposition Collapses: In bad-faith litigation, when defense counsel cannot produce an adjuster who personally reviewed the medical evidence and exercised human judgment, the insurer has virtually no defense against claims of reckless claim handling (*Egan v. Mutual of Omaha*).
The Watchdog Scorecard: How to Identify High-Risk AI Claims Vendors
Claims executives and compliance officers should immediately scrutinize vendors displaying any of the following red flags:
- Red Flag 1: The “Instant Settlement Offer” Claim: Any vendor claiming their software can formulate a settlement offer without human examiner intervention is marketing an unlicensed adjusting engine.
- Red Flag 2: Black-Box Liability Apportionment: Vendors claiming an LLM can “read a police report and assign fault percentages” are usurping a discretionary legal evaluation reserved strictly for licensed professionals.
- Red Flag 3: Synthetic General Damage Calculations: Algorithms that output an ungrounded “pain and suffering score” without citing specific, verified clinical records in the file violate state fair-settlement statutes requiring individualized factual justification.
What Compliant Architecture Actually Looks Like
Technology does have an indispensable role to play in modern claims handling—but as examiner-controlled decision support, not autonomous replacement. Compliant tools accelerate clerical indexing, organize treatment timelines, extract itemized charges, and tie every single finding directly to an underlying Bates-numbered document page, while preserving 100% of the discretionary adjusting authority for licensed examiners.