\n\n
Key developments:
- Major carriers are beginning to embed AI exclusions in commercial general liability (CGL) policies after the Insurance Services Office released model language in January, with brokers like Arthur J. Gallagher noting the language is "beginning to appear" on CGL contracts and specifically bars coverage for generative‑AI tools used in construction activities.
- Despite these moves, some insurers such as AXA XL are still hesitant to apply blanket AI exclusions, with its chief underwriting officer confirming they are "not applying AI exclusions" yet.
- The pace of formal exclusions is rising: ScienceSoft reports that over 60 property‑and‑casualty groups have filed AI‑related exclusions with regulators as of July 2026.
- At the same time, dedicated AI‑liability products are emerging. Munich Re’s aiSure suite and Armilla AI’s standalone AI liability policy (now up to $25 million) provide affirmative coverage for model errors, inaccurate outputs, and autonomous agent actions.
- New market infrastructure is being financed: the Artificial Intelligence Underwriting Council (AIUC) secured a $40 million Series A round to build insurance and certification standards for AI agents, signalling growing investor confidence in a separate AI‑insurance ecosystem.
- Regulators are moving toward oversight: the NAIC is finalising its AI Risk Evaluation Supplement to give state regulators a tool for assessing insurers’ AI governance, while state filings in Puerto Rico and Arizona are already collecting detailed AI usage data.
Why it matters:
- The emergence of exclusions signals that insurers see AI as a distinct, potentially unquantifiable source of loss, and they are beginning to protect their balance sheets by carving out those risks.
- Parallel development of dedicated AI‑liability policies shows the market is also creating supply‑side solutions for firms that need coverage for AI‑related errors, reducing the “silent AI” gap where existing policies may not respond.
- Funding for AI underwriting standards (AIUC) and the launch of specialized products (aiSure, Armilla) indicate a maturing ecosystem that could eventually separate AI risk from traditional lines, improving pricing accuracy and claim handling.
- Regulatory tools like the NAIC’s AI supplement will push insurers toward transparent AI governance, potentially limiting opaque exposures and fostering industry‑wide best practices.
What to watch next:
- Adoption rates of the ISO AI exclusion language across CGL and professional lines as renewal cycles begin later in 2026.
- Growth of dedicated AI‑liability capacity, especially as Munich Re and Armilla expand limits beyond $25 million.
- Implementation of the NAIC AI Risk Evaluation Supplement and state‑level AI bulletins (e.g., Arizona, New York’s RAISE Act) that could mandate human oversight of AI decisions.
- Further funding rounds for AI underwriting councils and insurtech firms building AI‑centric platforms, which may accelerate the shift from "silent AI" coverage to explicit endorsements.
- Real‑world claim data from early AI‑related losses (e.g., the Hugging Face breach) that will inform underwriting models and may trigger broader policy changes. [[1, 2, 3, 4, 5]]