The AI Exclusion Just Showed Up in a D&O Form
Our August 12 issue told readers not to assume the general liability AI-exclusion forms making the rounds, CG 40 47, CG 40 48, and CG 35 08, say anything about D&O, E&O, or fiduciary liability coverage. Those needed their own separate check. That check now has something concrete to check against. Coverage counsel have published the full text of what's described as the first "absolute" AI exclusion drafted for management liability lines, introduced in 2025 and intended for use in directors and officers, errors and omissions, and fiduciary liability products, attributed to W. R. Berkley (Hunton Andrews Kurth). Read it carefully, because it's broader than the summary suggests. It excludes claims arising from the use, deployment, or development of AI by any person or entity, not just the insured, including AI-generated content, an insured's failure to detect content created through a third party's use of AI, and an insured's own inadequate or deficient AI policies, practices, procedures, or training. It also reaches statements and disclosures about AI, violations of AI laws and AI-disclosure rules, and any regulatory demand that the company investigate or respond to AI risks.
Two things worth keeping straight. First, this isn't a carrier announcement. Berkley couldn't be reached for comment on the trade coverage, didn't respond to Reuters, and one version of the story hedges the description with "reportedly" (Insurance Journal). Reuters described the language as recently drafted and quoted coverage counsel saying it "may never see the light of day commercially" (Reuters). Second, it isn't the only such form circulating. Hamilton Select and Philadelphia Indemnity generative-AI exclusions surfaced in the same window, and Cincinnati Financial and Frederick Mutual have separately explored AI-restricting wording; broker pushback has already forced one U.S. carrier to withdraw AI exclusions it tried to introduce (The Insurer).
None of that makes this a new category of exposure. It's the D&O/E&O gap this newsletter already flagged, now with published language to hold up against your own policy instead of a hypothetical to go verify. What makes it worth your attention isn't the headline, it's the definition. "Artificial Intelligence" is defined as any machine-based system that infers from input how to generate predictions, content, recommendations, or decisions, not just generative AI. Read literally, that reaches tools your organization has used for years and never thought to call AI (Jones Day).
Think back to your last renewal call. Your broker walked the tower line by line, someone asked about AI, and the answer was that the general liability forms were handled. Everyone moved to the next item. Nobody in that room was thinking about the director who signed an AI vendor contract two quarters ago, or the CFO who approved an AI-driven cost cut. Under this language, they're the ones holding an uninsured claim. Not the company. Them.
If you ran the audit this newsletter suggested two issues ago and your broker's answer was "the GL forms don't apply here, you're fine," that answer was incomplete. The published language is proof the D&O, E&O, and fiduciary liability lines need their own separate, documented answer, not one borrowed from the general liability conversation. Ask about attachment, not just availability: the ISO general liability endorsements are optional forms that cover generative AI only on CGL and products/completed-operations lines (Insurance Journal), and the management liability form's commercial deployment hasn't been confirmed by any carrier.
This month's action: if you haven't already gotten a written answer to the D&O/E&O question this newsletter raised on August 12, get it now. Ask two specific questions: does any AI exclusion attach to our management liability tower, and if so, on what form? And does its definition of AI reach predictive or machine-learning tools, or only generative AI? The second question is the one most boards aren't asking.