86% want people — not the model — on the biggest AI calls
Version 5.0 of the AI Risk Evaluation Supplement, formerly the AI Systems Evaluation Tool, was open for 30 days of public comment ending September 29.
Know whose login that portal uses
New Jersey bill would fine insurers $5,000 for AI claim denials.
A new market for insuring AI systems just got $40 million.
An insurer launched AI underwriting. Does it reach US agencies?
Marsh says AI could cut London market placement time to hours.
New York created an AI office, not an insurance rule.
One AI insurer now writes small commercial in 43 states.
Most businesses worry AI will make cyberattacks worse.
Vertafore says its AI cut document review time 90%.
A vendor made a top-50 list. That's not a client.
An AI company just bought an insurance agency in Florida.
Every AI deployment story this month came from the company deploying it. LUBA, Acrisure, Mosaic, VIPR, Patra, every one a company release or a company statement
A California court fined a lawyer $999.99 this week for citing AI-hallucinated cases in a lawsuit involving State Farm.
Mosaic's answer to endless AI pilots: put the engineers in the underwriting workflow. An AI Lab and a Forward Deployed Engineer program, with engineers beside u
Patra's new AI platform now handles certificate of insurance delivery after bind, check it before it leaves your desk anyway.
Survey: most insurers would trust AI only on repeatable work
Survey: 13% of insurers say their AI operating model is advanced
Audit your submissions before your carrier's model does. Pull your last ten. Count how many had complete loss runs, correct entity names and values on day one.
ISO's generative AI exclusion is already on 2,369 CGL filings
Most agencies can't name every AI tool touching a client file, and an E&O renewal application can ask cold.
Covered entities must register with the state's DIGIT program and file a disclosure statement at least every other year starting January 1, 2027.
The Big Data and AI Working Group plans to expose an updated version for comment before a possible Fall 2026 adoption vote.
A5494 would let insurers use AI for triage, fraud scoring and approvals, but not the final decision to deny a homeowners, auto or flood claim.
A newswire release describes what it calls the first federal ruling on whether AI-assisted workplace investigations are protected by attorney-client privilege,
A RAND Corporation study on AI's insurability names the carrier's exclusion language, which reaches any actual or alleged use, deployment or development of AI.
Texas's TDI Bulletin B-0003-26 clarified that AI-enabled insurance activities remain subject to existing law, Risk & Insurance reports.
RAND recommends a standardized AI Coverage Notice so carriers state whether AI-related losses are covered, excluded or silent by line, Insurance Business report
The September 10 industry letter says significant AI-related developments can count as material changes to cyber risk under Part 500.
The expanded program will look at how insurers use machine learning in underwriting, pricing and claims.
Among them: no claim can be down-coded or a payment reduced by AI alone.
The Big Data and AI (H) Working Group opened a 30-day comment period on August 31, with adoption in view for the fall meeting.
A California judge just fined a lawyer $999.99 for filing AI-generated citations to cases that don't exist.
What happens to your submission in the ninety seconds after you hit send — as the vendors describe their own tools. A thin submission does not get a slow answer
The AI claim-denial cases prove less than the headlines say. The "90% error rate" is an allegation. The "1.2 seconds of review" is an allegation. In one case th
Every AI deployment story this month came from the company deploying it. LUBA, Acrisure, Mosaic, VIPR, Patra, every one a company release or a company statement
AI liability is a coverage question, not a product. Two families of AI exclusion are on the market and they do not define AI the same way. Berkley's management
The fear is that AI replaces underwriters. What is actually moving is triage. Small, clean accounts auto-clear; everything else routes to a human with a smaller
Three questions for your underwriter this renewal, and the reason to ask them before the quote, not after the declination.
Rewrite your cover letter for a machine and a human. The first paragraph is now read by a model before an underwriter sees it.
Audit your submissions before your carrier's model does. Pull your last ten. Count how many had complete loss runs, correct entity names and values on day one.
The vendor demo question that ends the meeting early: "Name a customer who has been live twelve months, and tell me what broke." A real deployment has a broken
Acrisure's AI announcement is about connecting the brokerage, not adding a chatbot. Auris AI, built on Palantir, pulls client, carrier, contract and operational
A regional workers' comp carrier is live on AI submission intake. LUBA Workers' Comp, Baton Rouge, on Convr's underwriting workbench: intake and triage reading
Mosaic's answer to endless AI pilots: put the engineers in the underwriting workflow. An AI Lab and a Forward Deployed Engineer program, with engineers beside u
The less flashy side of insurance AI: making risk, premium and claims files usable. Bridgepoint took a majority stake in VIPR, the delegated-authority data plat
What does "95% accuracy" actually count? Ten quantified claims from submission-AI vendors, quoted as written, and not one of them publishes the denominator. Two
A California court fined a lawyer $999.99 this week for citing AI-hallucinated cases in a lawsuit involving State Farm.
'AI-native core system' does not mean AI is deciding your risk yet.
Patra's new AI platform now handles certificate of insurance delivery after bind, check it before it leaves your desk anyway.
Four insurance AI platforms launched in the same eight-day window this month, none of them published a number.
A Business Wire release built around an ISG survey says 83% of the global insurance market would hand repeatable work to AI, with 75% wanting an insurance-speci
Most agencies can't name every AI tool touching a client file, and an E&O renewal application can ask cold.
Your comparative rater suggests a carrier. You bind a different one. Six months later a client asks why, and the file has no answer.
Keep a log of how long each market takes to respond. It is the only baseline you control.
A funding round is not evidence that an underwriting tool works. Beagle Labs raised $4.1 million, and its release names no insurer or MGA using the platform.
Make the next AI decision with better evidence.
Join the weekly Renewal Report for source-backed developments, practical checks, and clear limits on what is known.
Get the weekly reportOne email a week. Unsubscribe at any time.