AI in underwriting, pricing, and claims — classified as access-to-essential-services AI under the EU AI Act
Discriminatory pricing, opaque claims denial, and proxy variables in insurance AI are the patterns regulators and plaintiffs are building cases around. The early-mover audit advantage is still available — but the window is narrowing.
AI systems in scope for Insurance
Every system below is covered in a standard iDharma engagement. Complex or multi-system deployments are scoped on request.
Risk classification, eligibility decisions, and pricing models — proxy discrimination testing, explainability, and Annex III conformity documentation.
Automated claims approval, denial, and fraud-scoring models — accuracy benchmarking, false-positive analysis, and human oversight verification.
Dynamic and personalised pricing AI — differential pricing by protected class, regulatory rate-filing alignment, and proxy variable audit.
Models flagging claims for investigation — demographic performance disparities, accuracy on sub-groups, and documentation of the escalation process.
What we audit against
Every iDharma Insurance engagement maps simultaneously against the frameworks below — producing one gap register, not three separate reports.
Insurance AI used in underwriting, pricing, and claims is classified as high-risk access-to-essential-services AI. Conformity assessment, human oversight mechanisms, and technical documentation are mandatory.
Requires insurers to assess AI and external data sources for unfair discrimination. Applies to all personal lines insurers writing NY business.
Prohibits insurers from using external data, algorithms, or predictive models that unfairly discriminate on the basis of race, colour, national or ethnic origin, religion, or sex.
How an iDharma audit works in Insurance
We do not accept vendor documentation as evidence. We do not produce checkbox compliance reports. Every audit produces a named auditor, a cited methodology, and a straight answer on exactly where your AI stands — and what to fix first.
See how we work →We test every underwriting and pricing feature for proxy discrimination — including geodemographic proxies, behavioural data, and third-party data sources.
We evaluate claims AI for demographic performance disparities — a fraud model with a higher false-positive rate for one demographic is a regulatory liability, not just a model quality issue.
We verify that "the vendor handles compliance" is not being treated as a defence — it has not succeeded in any EU enforcement action we have reviewed.
We produce a conformity assessment package structured for the EU AI Act Annex III requirements — so you are not scrambling to produce documentation when the first enforcement action is announced in your sector.
Explore other domains
Using AI in underwriting, pricing, or claims?
The enforcement window for insurance AI is still open. Start with the free Risk Snapshot to understand your exposure before the first major fine is issued.