What Happened
In January 2026, Verisk’s ISO Core Lines Services made a series of new endorsement forms available to carriers across the United States. Among them are several forms specifically designed to exclude generative artificial intelligence exposures from commercial general liability policies.
The key forms are:
| Form Number | Coverage Affected | What It Excludes |
|---|---|---|
| CG 40 47 01 26 | CGL Coverage Part | Bodily injury, property damage, and personal/advertising injury arising out of generative AI |
| CG 40 48 01 26 | CGL Coverage B only | Personal and advertising injury arising out of generative AI |
| CG 35 08 01 26 | Products/Completed Operations | Bodily injury and property damage arising out of generative AI |
These endorsements are optional. Carriers choose whether to adopt them. But Verisk has reported strong interest from carriers, and multiple industry sources anticipate rapid adoption. At least half a dozen insurers had already filed to adopt the new wordings shortly after Verisk’s announcement. Separately, carriers including WR Berkley, Cincinnati Financial, Frederick Mutual, and Philadelphia Insurance have independently filed their own AI exclusion language with state regulators.
The Exact Language
All three forms use an identical definition of generative AI:
“Generative artificial intelligence” means a machine-based learning system or model that is trained on data with the ability to create content or responses, including but not limited to text, images, audio, video or code.
Verisk based this definition on existing definitions from government sources and the National Association of Insurance Commissioners (NAIC).
The exclusion language in CG 40 48 (Coverage B) reads:
This insurance does not apply to: “Personal and advertising injury” arising out of “generative artificial intelligence”.
The critical phrase is “arising out of.” In insurance law, “arising out of” is interpreted broadly. It does not require direct causation. A claim needs only a causal connection to generative AI to trigger the exclusion. This means a wide range of AI-adjacent claims could be denied.
Why This Matters More Than Most Brokers Realize
The impact extends well beyond companies that are obviously “AI companies.” Consider how broadly generative AI has penetrated enterprise operations:
71% of surveyed businesses have implemented generative AI in at least one function (Geneva Association, 2025).
57% of companies already have AI agents running in production environments (G2, August 2025).
29% of employees have turned to unsanctioned AI agents for work tasks, and the average enterprise has an estimated 1,200 unofficial AI applications in use (Microsoft Data Security Index, 2026).
By 2028, Gartner projects 90% of B2B buying will be AI-agent-intermediated.
This means your commercial clients are almost certainly using generative AI somewhere in their operations — even if they don’t know it. Customer service chatbots, marketing content generators, AI-assisted code review, automated email drafting, AI-powered CRM features — all of these fall under the ISO definition of “generative artificial intelligence.”
The practical consequence: a client using Salesforce Einstein, HubSpot AI, Zendesk AI, or any of hundreds of enterprise tools with embedded generative AI features could face claim denials under these exclusions. The exclusion does not distinguish between an enterprise that built its own AI system and one that simply uses a vendor product with AI features embedded in it.
The Broader Exclusion Landscape
Verisk’s forms are the standardized version, but they are not the only — or even the broadest — AI exclusions in the market.
WR Berkley has proposed an “absolute exclusion” that would bar claims tied to “any actual or alleged use” of AI, even if the technology forms only a minor part of a product or workflow. This language goes significantly further than Verisk’s generative-AI-specific definition, potentially covering traditional machine learning, rules-based automation, and any system that could be characterized as “artificial intelligence.”
AIG has told regulators it has “no plans to implement” its proposed exclusions immediately, but wants the option available as the frequency and scale of claims increase.
These exclusions are also appearing beyond CGL policies. AI exclusion language is being introduced in Directors & Officers (D&O), Errors & Omissions (E&O), and professional liability policies. As Harvard Law School Forum on Corporate Governance has noted, most of these exclusions are “near absolute in scope, precluding coverage in full for any claim in any way related, directly or indirectly to the usage of any AI.”
What Brokers Should Do Now
1. Audit your clients’ policy language
Before the next renewal, review every active policy for every commercial client. Search for the terms “Artificial Intelligence,” “Algorithm,” “Machine Learning,” “Automated Decision,” “Generative AI,” and “Neural Network.” Document which policies contain AI-related exclusions and which are still silent on AI. Policies renewing in Q1 and Q2 2026 are the most immediately affected.
2. Inventory your clients’ AI usage
Most enterprises do not have a complete picture of their own AI deployment. Ask each client’s IT and operations teams to survey departmental AI tool usage, including marketing tools (content creation, social media), customer service platforms (chatbots, helpdesk), HR systems (applicant tracking, resume screening), financial tools (automated reporting, expense management), and development tools (AI-assisted coding, testing).
3. Negotiate exclusion narrowing
Even when carriers want to add AI exclusions, negotiation room often exists. Request defined terms that limit what counts as “AI” to avoid overly broad definitions. Seek carve-backs for governed AI use where the client can demonstrate strong controls — documented policies, human oversight, access restrictions, and monitoring. Get written examples of excluded versus covered scenarios. Clients with documented AI governance frameworks are in a significantly stronger negotiating position.
4. Consider the role of independent risk assessment
Carriers are more likely to grant narrow exclusions or carve-backs when presented with evidence that a client has conducted a formal technical assessment of their AI deployments. This includes documentation of what AI tools are in use, what data they access, what guardrails are in place, and what failure modes have been tested for. A third-party risk assessment report provides the kind of evidence that supports negotiations with underwriters who are uncertain about AI exposure.
5. Explore emerging specialty coverage
The same market forces driving exclusions are also creating new specialty products. Testudo, an MGA launching in early 2026, is specifically designed to insure AI liabilities for U.S. mid-market enterprises. Armilla AI offers performance warranty products backed by Swiss Re, Greenlight Re, and Chaucer. Munich Re’s aiSure has offered AI performance coverage since 2018. These products may fill gaps that traditional carriers are creating, but coverage scope, pricing, and governance requirements vary significantly.
The Parallel to Silent Cyber
This transition closely mirrors the “silent cyber” crisis of 2015–2023, when insurers discovered they were inadvertently covering cyber losses under property and casualty policies that were never designed to handle them. The insurance industry’s response was to introduce explicit cyber exclusions and create a standalone cyber insurance market.
The same pattern is now repeating with AI. Silent AI coverage — where AI risks were neither explicitly included nor excluded — is disappearing. Explicit exclusions are being standardized. Standalone AI coverage is emerging as a separate product category. Industry analysts expect AI exclusions to become standard across most policies by late 2027.
Brokers who prepared their clients for the silent cyber transition avoided coverage gaps and claim denials. Those who ignored the warnings faced expensive lessons. The window to prepare for the AI transition is now, while negotiation leverage still exists and before exclusions become fully standardized.
Key Takeaway
The era of AI coverage by default has ended. Your commercial clients are using AI whether they realize it or not. Their existing policies are being modified to exclude AI risks, often with broad language that could deny claims with even a tangential connection to AI. The combination of increasing AI usage and decreasing AI coverage creates a widening exposure gap. Proactive assessment, documentation, and negotiation — conducted before renewal — is the only way to protect your clients and your book of business.
Sources: Verisk Core Lines Services, ISO Forms CG 40 47, CG 40 48, CG 35 08 (January 2026 edition). Geneva Association, “Gen AI Risks for Businesses” (2025). G2 Enterprise AI Agents Report (2025). Harvard Law School Forum on Corporate Governance, “The Hidden C-Suite Risk of AI Failures” (September 2025). Bloomberg Law, “Policyholders Should Negotiate to Limit AI Exclusions at Renewal” (December 2025). Insurance Business Magazine (2025–2026). Financial Times / TechCrunch reporting on AIG, WR Berkley, and Great American filings (November 2025). Microsoft Data Security Index (2026). Gartner Strategic Predictions for 2026.
Published by Gridex Inc. · GDX-BR-001 · March 2026
This briefing is provided for informational purposes only and does not constitute legal, insurance, or financial advice.