Practice Area · AI in Actuarial Science

AI in Insurance

Daily intelligence on AI deployment in insurance — carrier production use, vendor accountability, model governance, and regulation shaping build-vs-buy decisions.

From Our Desk Original analysis · AI

All AI insights →
AI in Insurance

TabPFN Trails the GLM on Motor Frequency in New Pricing Tests

A Poisson GLM beat the pretrained TabPFN model on claim frequency in both motor portfolios tested. Language-model embeddings helped a GLM most on small books, leading it by about 6% at 10,000 policies.

AI in InsuranceNov 12 · 5 min

Cowbell Launches a Standalone Liability Policy for AI Agents

Cowbell's standalone AI liability policy pays a small business's own losses and third-party claims from one AI Incident trigger. It goes on sale January 1, 2027, through wholesale brokers on a non-admitted basis.

AI in InsuranceNov 11 · 5 min

Model Risk Management for Actuarial Models, From GLM to LLM

Bank supervisors define validation as conceptual soundness, ongoing monitoring and outcomes analysis. A GLM and a GBM pass all three by back-testing. An LLM-assisted process has no back-test, only sampled human review.

AI RegulationNov 10 · 6 min

NAIC Puts AI Supplement v7.0 to a Vote With GLMs Still in Scope

The NAIC's Big Data and AI working group votes on 16 November in Dallas. Version 7.0 keeps GLMs in Exhibit A but lets regulators exclude them from the whole supplement.

AI in InsuranceNov 9 · 6 min

Insurers Won 25 AI Patents in October and the Year Is Exactly Flat

Tracked insurance filers received 25 granted US AI patents in October 2026, against 22 a year earlier, and 35 of their pending AI applications were published. State Farm filed 18 of the 35.

AI in InsuranceNov 5 · 4 min

Intact Puts AI Pricing Earnings at C$250M and Says It Will Double

Intact said machine learning prices 94% of its Canadian personal lines and earns close to C$250 million a year before tax, a figure it expects to double, as catastrophe losses pushed its combined ratio to 94.0%.

AI UnderwritingNov 4 · 5 min

SOA Study Cuts Underwriting Bias Gap to 0.071 for 0.001 of AUC

An SOA Research Institute simulation puts a GLM and an adversarial neural network 0.001 of AUC apart, while their true-positive-rate gaps at a 2% referral cutoff are 0.114 and 0.071.

AI Patent Watch AI patent grants by filer · 12mo to Nov 2026

Patent coverage →
12mo to Nov 2026 292 AI patent grants across 33 insurance-sector applicants in the trailing 12 months. Flat against the prior 12 months. UnitedHealth leads with 74.

Granted US utility patents in the patent office's artificial intelligence and machine learning classifications, by filing organization. Source: United States Patent and Trademark Office, as of November 15, 2026. ▲ marks more grants than the prior 12-month period, a measure of activity rather than of quality. Full leaderboard, quarterly trend and the newest grants: AI Patent Watch.

Standards & Society SOA, CAS, and regulator activity on AI

All AI standards →

The Daily Actuary, in your inbox

One short email each weekday: today's top original stories and the wire by line of business, with our 'Why It Matters' read on each.