Insurers

Insurance regulators were among the first to write rules specifically about artificial intelligence. New York’s Circular Letter No. 7 asks insurers to show that external data and AI systems do not produce unfair discrimination and that governance is real. The NAIC’s model bulletin is being adopted state by state. In Canada, the CCIR-CISRO fair treatment guidance and the AMF’s guideline on the use of AI set the expectation, and FSRA has proposed an operational risk framework for auto rating and underwriting. In Europe, EIOPA’s opinion tells national supervisors what to look for.

Conditionary’s financial-services pack carries those instruments alongside 23 NYCRR Part 500, the AMF’s model risk and ICT risk guidelines, provincial privacy law and the general consumer protection rules. The set-up guide asks about underwriting and pricing models, claims triage and fraud referral, and chat assistants, and the controls that an actuarial and model risk function already runs are proposed for confirmation.

Two things make this pack unusual. First, the insurer evidence summary was written with the insurer as the reader, which means an insurer can also use it to look at itself. Second, case checks test the per-case duties, such as whether every adverse underwriting action carried its explanation within the time your own service standard sets.

Ready to run: Run the Evidence summary for an insurer against your own book, then ask your commercial policyholders for the same thing. See the reports.