Learning Center · AI and insurance
AI and insurance, in numbers.
How much AI risk sits silently in existing policies, the new exclusions and AI covers, the claims already in court, the questions carriers now ask, and the rules for insurers. Every figure links to its primary source.
Silent AI
Most AI risk sits in policies that never mention it.
of AI coverage falls under “Silent AI” insurance, as of March 2026.
Aon AI Fact Sheet 2026, p. 1potential insurance lines that may respond to more than a dozen AI perils, in a figure RAND cites.
RAND, September 2026The market moves
Exclusions on one side. Affirmative AI cover on the other.
proposed effective date of the optional generative AI endorsements in the ISO multistate general liability filing. Adoption varies by state and carrier.
Verisk, July 2025 · Our research note on CG 40 47Beazley launched an AI Voluntary Shutdown solution: protection for organisations that suspend their own malfunctioning AI systems.
BeazleyHSB introduced AI liability coverage that protects small and medium-sized businesses from lawsuits.
HSB, Munich ReCFC added enhanced affirmative AI cover to its cyber policy, “to provide clarity and confidence”, and extended business interruption from 12 to 18 months.
CFCAI use is the top medium-term risk for corporate risk managers, and enters their near-term top five for the first time.
FERMA and PwC Global Risk Manager Survey 2026managing-agent responses to a Lloyd’s market survey of AI loss scenarios. Underwriters rated professional indemnity as the highest potential impact.
Lloyd’s Market Association, January 2026AI and big data rank first among experts in AXA’s Future Risks Report, ahead of climate change, for the first time since the report was launched. AI ranked fourth in 2025.
AXA Future Risks Report 2026, with Ipsos, October 2026A frontier AI developer told investors that no single insurer, and possibly not the insurance industry as a whole, may have the capacity to underwrite cover for a catastrophic AI incident.
Anthropic, Form S-1, 26 October 2026In court
AI claims are already in the securities courts.
AI-related securities class actions filed in the first half of 2026.
Cornerstone Research, July 2026disclosure dollar loss from those filings: 73% of the half-year total, from 13% of core filings.
Cornerstone Research, July 2026What carriers ask
The questions on the application are changing.
- “Does a human need to verify accuracy before AI takes action?”A sample carrier application question, Aon AI Fact Sheet 2026, p. 6.
- “Are AI actions logged so potential errors can be reviewed and remediated?”A sample carrier application question, Aon AI Fact Sheet 2026, p. 6.
Aon AI Fact Sheet 2026, published as of 1 August 2026.
At claim, four questions decide an AI agent loss.
- When did the AI first act outside its authority?
- Who stopped it, and when?
- Did they hold the authority?
- What ran afterwards?
The rules for insurers
Regulators already expect a written programme.
jurisdictions have adopted the NAIC AI Model Bulletin, 25 states and the District of Columbia, as of 31 August 2026. California, Colorado, New York and Texas have their own insurance-specific rules.
NAIC adoption map · The AI rules for insurersNext
A driving record, for AI.
Figures are as each publisher states them. Regulayer™ supplies the evidence; the insurer makes the determination.
