AI Liability Insurance for Law Firms

What AI liability insurance is

AI liability insurance covers a risk that lacks a clean home in the existing policy stack: financial liability when a regulated professional uses artificial intelligence in client-facing work. For licensed attorneys, the risk lands first against the lawyers professional liability (LPL) policy. That is the malpractice-cover form most US firms carry. Whether the LPL policy actually pays an AI-driven claim is the load-bearing question. In 2026, the answer is bifurcated.

Two related cover concepts are emerging. The first is silent AI cover: an existing professional liability or errors and omissions form has not been amended for AI, so the carrier adjudicates an AI-related claim under pre-existing policy language. The second is affirmative AI cover: a separate product or endorsement that explicitly covers AI-related claims. Affirmative cover typically conditions on documented governance practices and an AI risk assessment.

Through 2024 and 2025, almost all US law firms operated under silent AI cover by default. By the end of 2025, that silence began to disappear from renewal documentation. It was replaced by some combination of underwriter questions, governance-document requests, manuscript exclusions, and a small affirmative-coverage market. Sections below describe what has been filed publicly, what remains unannounced, and how a firm can reduce exposure on either side at renewal.

Does my LPL policy cover AI mistakes?

The honest answer: "it depends on your policy and your carrier, and the market is moving against you." WTW's Insurance Marketplace Realities 2026, Errors and Omissions outlook (October 2025), describes January 2025 to January 2026 as a "structural break" in the professional liability market. Silence on AI began to disappear from renewal documentation. It was replaced by affirmative-but-conditioned coverage, sweeping exclusions, or governance-contingent silent cover.

For a firm at renewal, the assumption "I have LPL, so AI is covered" is no longer a reliable default. Suppose a claim alleges that a lawyer cited a hallucinated case to a tribunal. Or suppose confidential client information was disclosed through a self-learning tool. The carrier will analyze the claim against the actual policy form on the date of the alleged act. That form may be silent on AI, may have an exclusion endorsement attached, or may have an affirmative coverage grant. Three carriers writing the same LPL line can land in three different places.

Practical evidence that the bifurcation is real, not theoretical, sits in the questions underwriters now ask at renewal. Stan Sterna is Aon's senior vice president and risk control lead for accounting and law firm professional services. He summarized the posture in April 2026:

"They're asking firms, 'Do you use AI? Do you police it? Do you have protocols in place?'"
Stan Sterna, Aon, in Digital Insurance / Accounting Today, April 2026.

That set of questions is now standard at most professional liability renewal cycles. Sterna also noted that cyber underwriters are several years ahead of professional liability underwriters on AI. The framing is useful: cyber renewals already routinely require documented controls, and LPL renewals are converging toward the same posture.

Which carriers have filed AI exclusion endorsements?

As of May 2026, the headline is narrower than trade-press coverage suggests. No major US LPL writer (CNA, Travelers, Chubb, Hartford, Markel, ALPS, or any state mutual) has publicly filed an explicit AI exclusion endorsement on lawyers professional liability paper. Publicly filed AI exclusions sit on adjacent lines: management liability (D&O, EPL, Fiduciary) and general liability, not LPL.

The named filings most often cited at renewal:

• W. R. Berkley, Form PC 51380 00 (06-24), Artificial Intelligence Exclusion (Absolute), amending Berkley's management liability base policy family. Hosted publicly via Hunton's regulatory tracking. Does not on its face attach to LPL forms.
• Hamilton Select Insurance, generative AI exclusion on professional liability paper. Defines generative AI to include ChatGPT, Bard, Midjourney, and DALL-E by name.
• Verisk / ISO, CG 40 47, CG 40 48, CG 35 08, effective January 1, 2026, on general liability. Largest publicly visible AI form-filing event in the US market to date; template language is being ported to other lines.
• Unnamed LPL programs reported by independent brokers: a September 2024 Vanguard Specialty memo by Stephen Van Wert and an L Squared Insurance Agency posting describing manuscript LPL exclusions on unnamed programs. Treat as market signal, not a single carrier's policy.

Existing LPL exclusions can already deny coverage on AI facts in four ways: the professional services definition, intentional acts, breach of confidentiality, and unauthorized practice. ALPS' August 2025 framework sets out the analysis. For the form-by-form breakdown, plus FY2025 10-K language from Travelers, Chubb, Markel, CNA, Hanover, and Berkley, see the carrier renewal documentation page. The carrier tracker has the per-carrier published-position breakdown.

Which carriers offer affirmative AI coverage?

Affirmative AI coverage is a separate product or endorsement that explicitly covers AI-related claims. It sells alongside (not in place of) the firm's existing LPL policy. Carriers offering affirmative cover in 2026:

• Armilla AI with Lloyd's syndicate Chaucer (1084) and Axis Capital. Affirmative AI liability insurance with USD 25M limits, expanded February 10, 2026 into "Vanguard AI" bundling cyber, technology E&O, and AI. Lawyer hallucination cases explicitly named in scope.
• Counterpart, US managing general agent. Affirmative AI Coverage on miscellaneous professional liability with Tech E&O insuring agreement, expanded November 26, 2025.
• Coalition. AI Coverage product within its cyber program; most relevant for cyber-adjacent AI exposure.
• Munich Re aiSure. Performance-warranty product including aiSure-General Liability for bias and discrimination claims; primarily relevant for AI-developer exposure.
• Hiscox. Affirmative AI cover within its technology professional indemnity book.
• Beazley. Public posture is "no AI exclusion planned" (Bob Wice, February 2025), meaning silent AI cover by explicit choice, not by default.

These products share a common feature: they operate parallel to, not inside, traditional LPL. When a firm buys affirmative AI cover from Armilla, Counterpart, or Munich Re, it sits on top of the existing LPL policy, not as a substitute. Underwriting requires a documented AI risk assessment as a precondition. Documentation that satisfies an existing LPL underwriter at renewal generally satisfies the affirmative-cover carrier too, which is the operational case for documenting once.

The per-carrier breakdown, including Lloyd's capacity, named broker channels, and the LMA International Professional Indemnity Committee report (September 2025), is on the carrier renewal documentation page.

What to ask your underwriter at renewal

Sterna's framing translates into seven concrete questions for renewal. None are unusual. If any are absent from the conversation, that's the signal that the carrier has not yet formed a posture.

1. Is silent AI cover still in effect on this policy form, or has it been amended for the renewal? If "amended," ask for the endorsement language in writing before binding.
2. Has the carrier filed any AI-specific exclusion endorsement on this line, in this state? The answer may be "not on this line yet"; that answer is informative because it tells you the form-side trajectory.
3. If an AI exclusion is added at any future renewal, what is the scope? The Berkley form is "absolute": any use, any deployment, any development. Some carriers will offer narrower carve-outs (specific tools, specific use cases). The scope question matters because absolute exclusion is materially different from a tool-list exclusion.
4. Does the carrier require an AI risk assessment, governance documentation, or training records as a precondition for renewal? If yes, ask for the specific documents the carrier wants to see, not just the categories. The list is the renewal binder.
5. Are there premium adjustments tied to AI use? Some carriers are pricing AI-using firms differently. The pricing question often surfaces before the form-side question does.
6. If a claim involving AI arises, what notice triggers and timing requirements apply? AI-related claims may not fit cleanly into the policy's existing claim definition. Confirm the notice mechanics before a claim arises, not after.
7. Does the carrier offer affirmative AI cover as a separate product, and at what price point? Even if the answer is no, the conversation surfaces whether the carrier expects to.

A small number of carriers, particularly the state-affiliated mutuals, are ahead of the broader market on AI-specific risk-management content. Camille Stell at Lawyers Mutual of North Carolina (the LMLIC mutual that writes LPL for the NC bar) is one example. She has authored multiple practice-side AI articles since 2024 and published an AI Use Policy template in December 2025. If your carrier offers a published AI policy template, treat it as the starting point for the renewal conversation. The carrier has, in effect, told the firm what documentation it expects.

How to document AI use for renewal

Documentation a carrier wants at renewal in 2026 is not different in kind from what a regulator or a sanctioning court would want. ABA Formal Opinion 512 is the national baseline ethics opinion on generative AI for lawyers. It maps directly to nine renewal-readiness artifacts. A firm that can produce all nine has what the carrier is most likely to request, plus a defensible record if a claim arises.

• Written firm AI policy (Rules 5.1, 5.3): a current policy, dated within the last 12 months, naming approved tools, prohibited uses, and supervisory responsibility.
• Vendor due-diligence file (Rules 1.1, 1.6): for each approved tool, terms of use, privacy policy, data retention, and training-on-input behavior reviewed and recorded.
• Tool-specific informed-consent language (Rule 1.6): for any self-learning tool into which client information is input, client-facing consent language that meets Opinion 512's "not boiler-plate" standard.
• Training and CLE records (Rules 1.1, 5.1, 5.3): every attorney and staff member using an approved tool has completed tool-specific training, logged by date.
• Pre-filing verification log (Rule 3.3): any filing that used AI research assistance is logged with the citation-verification step documented by the reviewing attorney.
• Usage log (Rules 5.1, 5.3): cross-matter register of who used which AI tool on which matter.
• Incident response procedure (Rule 1.6): a written procedure for AI-related confidentiality incidents, including client notification and bar reporting.
• Quarterly attestation (Rule 5.1): managing-partner sign-off, every quarter, that the program is current.
• Renewal-ready audit report: one-page status memo summarizing the program, suitable as the cover document for the carrier submission.

Each artifact is available as a free template on this site. A firm that adopts all nine has the renewal binder. The Opinion 512 compliance guide maps each rule to the artifact that documents it. State-specific overlays appear in the state tracker; where state guidance is stricter than Opinion 512, the state rule controls.

What is changing in 2026 and 2027

Three forward indicators are worth tracking from a firm's perspective.

• Claim volume is climbing. Generative AI-related US lawsuits rose 978% from 2021 to 2025. More than 700 cumulative cases were reported as of early 2026. Gartner has forecast more than 2,000 "death by AI" legal claims by year-end 2026 across all lines.
• The first publicly reported LPL claim denial citing an AI exclusion has not yet landed. When it does, expect carrier renewal questionnaires to harden quickly. Document AI governance now to head off the questionnaire pressure when it hits.
• State regulatory action is uneven. The NAIC AI Model Bulletin (December 2023) and the NY DFS Circular Letter No. 7 govern insurers' own use of AI in underwriting and pricing, not exclusion language in professional lines. As of May 2026, no state insurance regulator has issued a circular letter specifically addressing AI exclusions in professional liability.

Three things follow for a firm planning 2026 and 2027 renewals. Assume silent AI cover continues to disappear from renewal documentation. Expect the affirmative-coverage market to expand and require documented governance as a precondition. Watch for the first publicly reported LPL claim denial as the trigger event for accelerated form-side change. Lloyd's market timing (Y-prefix bulletin status, LMA International PI Committee September 2025 report) is on the carrier renewal documentation page.

Primary sources

• Berkley Insurance Company, Form PC 51380 00 (06-24), Artificial Intelligence Exclusion (Absolute): primary source PDF.
• WTW, Insurance Marketplace Realities 2026, Errors and Omissions outlook (October 2025): wtwco.com.
• ABA Standing Committee on Ethics and Professional Responsibility, Formal Opinion 512 (July 29, 2024): americanbar.org (PDF).
• New York Department of Financial Services, Circular Letter No. 7 (2024), Use of AI in Insurance Underwriting and Pricing: dfs.ny.gov.
• Kennedys, "Silent AI cover: the unforeseen risks for insurers" (2025): kennedyslaw.com.
• L Squared Insurance Agency, "Be Careful of New Artificial Intelligence Attorney Malpractice Insurance Exclusion": l2insuranceagency.com.
• Lawyers Mutual of North Carolina (LMLIC), risk-management library: lawyersmutualnc.com.
• Armilla AI, Affirmative AI Liability Insurance: armilla.ai.
• Counterpart, AI coverage: yourcounterpart.com.
• Munich Re aiSure: munichre.com.

Last verified against primary sources: 2026-05-04.