Week in Brief: Taming AI, decoding AIG, BMS on M&A, medmal on thin ice
Here's the quick hit of what mattered this week for insurance pros. Four themes, clean takeaways, and next steps you can put to work.
1) Taming AI: the discrimination suit that could open the floodgates
A high-profile suit alleging AI-driven discrimination put every carrier on notice. The message is simple: if a model touches pricing, underwriting, claims, fraud, or marketing, treat it like a regulated product feature, not a shiny tool.
- Inventory every model in production. Document purpose, inputs, outputs, owners, and approval dates.
- Run bias testing by protected class where legally appropriate. Keep evidence. Update it quarterly.
- Lock down data lineage. If you can't trace the data, you can't defend the outcome.
- Stand up model governance: approval gates, challenger models, kill-switch criteria, and audit trails.
Helpful guardrails: the NAIC AI bulletin and the NIST AI Risk Management Framework align well with what regulators will ask for.
2) Decoding AIG: big-carrier signals worth reading
Large carriers set the tone on capital, reinsurance appetite, and pricing resolve. Watch the tea leaves, not the quotes: how they talk about casualty severity, cat load, reserve adds, and expense discipline tells you where rates and terms head next.
- Rate adequacy: track combined ratio ex-cat and loss trend commentary, not just headline CR.
- Reinsurance: note retentions, hours clauses, and limit bought. That flows straight into primary appetites.
- Capital: buybacks vs. growth signals if capacity will tighten or loosen by line.
3) BMS on M&A: consolidation keeps reshaping placement leverage
Broker M&A is still compressing market access into fewer hands. That can help on complex towers, but it can also reduce optionality for middle-market buyers if panels narrow.
- For carriers: maintain multi-broker distribution where possible; protect access to niche MGAs and facilities.
- For buyers: set service KPIs in the BOR, clarify data ownership, and run competitive marketing at renewal for key lines.
- For MGAs: sharpen differentiation (data, speed, specialist UW authority) to stay essential in consolidated chains.
4) Medmal on thin ice: severity keeps biting
Verdict severity and defense costs continue to pressure results even as frequency stays muted. Capacity is available, but it's picky and priced for pain.
- Underwriting: tighten venue and specialty selection; push for risk-management commitments at the group level.
- Claims: escalate earlier, reserve earlier, and deploy early resolution playbooks on high-exposure cases.
- Capital: consider quota share to stabilize volatility and structured reinsurance for earnings protection.
What to do before next quarter
- AI controls: publish a one-page Model Use Policy and appoint accountable owners for each high-impact model.
- Pricing: refresh severity trend picks and social inflation assumptions; document the rationale for rate filings.
- Reinsurance: pre-wire renewal narratives with updated cat views and attachment strategy-don't wait for the slip.
- Distribution: stress-test reliance on any single broker; add a secondary channel for key lines or segments.
- Claims: expand verdict analytics and tighten panel counsel scorecards tied to outcome and cost metrics.
Need practical AI upskilling for underwriting, claims, and compliance teams? Explore focused programs at Complete AI Training.
Your membership also unlocks:
