AI Exclusions in Business Liability Insurance 2026: What Small Businesses Must Check Before Renewal

If the chatbot on your website gives a customer dangerously wrong advice, or an AI writing tool publishes a defamatory sentence under your company's name, who actually pays — you, the AI vendor, or your insurer? Ten years ago the question barely existed. In 2026 it may decide whether your business survives a single claim.

For policies renewed in 2026, increasingly the answer is you. Three ISO exclusion endorsements — CG 40 47, CG 40 48, and CG 35 08 — now remove generative-AI claims from general liability cover, and filings to add them are accelerating. One insurer, Berkley, already applies an absolute AI exclusion across its management liability policies.

Last updated: October 2026

The legal environment is moving at a pace most owners have not tracked. A Gallagher analysis cited by Insurance Journal found AI-related lawsuits rose 978% between 2021 and 2025, and another 137% in 2024-2025 alone — with claim mixes spanning patent infringement at 11.9%, copyright at 11.2%, and personal-injury-type claims at 10.2%. Insurers read those numbers and rewrote their forms.

This guide walks through what the new policy wording actually says, in plain English, question by question. It is not legal advice — policy language and state rules vary, so confirm your own position with your broker or counsel before acting on anything here. And the squeeze is broader than insurance: 2026 piled cost shocks on small businesses from several directions at once, from SaaS pricing to cover narrowing — Intuit's QuickBooks, for instance, raised three of its four plans for August 2026 renewals.

Source: Wikimedia Commons — AI Humans and Robots (CC BY 2.0)

Probably not, if your policy renewed in 2026 with one of the new standard exclusions attached — and you may not know unless you read the endorsement pages. Policies written a few years ago were usually silent on AI, which meant an arguable gap for both sides. Insurers are closing that ambiguity in their own favor, form by form. If you cannot find the words "artificial intelligence" anywhere in your wording, do not assume you are covered; assume the question is unresolved and get the answer in writing before renewal.

The scale of the rewrite is easy to underestimate. The Gallagher figures above — a near ten-fold rise in AI litigation in four years — turned AI from a hypothetical risk into a rated one. Attorney Alana McMullin of Lathrop GPM, quoted in Insurance Journal's August 2026 feature, said plainly that "these ISO exclusions were the spark of this AI exclusion boom." When the analyst community and the litigation data agree, wording changes follow at speed.

What Is the ISO AI Exclusion Endorsement CG 40 47?

CG 40 47 is a standard-form endorsement that insurance carriers attach to commercial general liability policies to strip out coverage for bodily injury, property damage, and personal and advertising injury arising from generative AI. It is one of three matching forms filed by ISO, the organization that drafts the standard wording most US commercial policies are built on. Wherever your policy uses ISO forms, these endorsements can be added at renewal with a signature.

ISO form	Where it applies	What it removes
CG 40 47	Commercial general liability	Bodily injury, property damage, and personal and advertising injury arising from generative AI
CG 40 48	CGL Coverage B only	Personal and advertising injury arising from generative AI
CG 35 08	Products and completed operations	Bodily injury and property damage arising from generative AI

Two details in the fine print matter more than the headlines. All three forms include a definition of "generative artificial intelligence", which pins the exclusion to generative systems specifically rather than all software — a meaningful line for businesses using predictive analytics rather than text generators. And the trio's breadth means an AI-related claim can fall through different doors depending on whether it is framed as injury, defamation, or a defective product output.

The insurers' side of the story is not simply risk aversion. Joe Lam, a Verisk liability vice president who helped write the endorsements, told Insurance Journal that without exclusions carriers "might just walk away from the risk" entirely — meaning the alternative to exclusion may be no cover at all. McMullin expects "rapid adoption of some version of an AI exclusion in most lines," though some carriers may instead keep AI risk and price it into premium.

Source: Wikimedia Commons — Artificial Intelligence and Robotics exhibition at the Heinz Nixdorf MuseumsForum (CC BY-SA 4.0)

What Is Silent AI Coverage, and Is It Disappearing?

Yes — and quickly. "Silent AI coverage" describes older policies that neither exclude nor explicitly affirm AI-related claims, leaving courts to guess intent. Since late 2025, insurers have been replacing silence with explicit wording, and regulators have been approving the new forms. If your renewal date falls in 2026, you are renewing into that transition.

The reported timeline tells the story. The Financial Times reported insurers retreating from AI cover in November 2025. Analysts at Evercore ISI said in December 2025 they expected generative-AI liability exclusions to spread. By February 2026, trade coverage reported US insurers adding generative-AI exclusions as regulators approved the forms. In April 2026, The Information reported that Berkshire Hathaway and Chubb had won approval to drop AI coverage. By July 2026, Bloomberg Law was reporting policyholder alarm over coverage gaps, and Claims Journal that insurer interest in AI exclusions was still growing.

Source: Barclay Damon LLP — official YouTube channel

For a small business, silence was never a strategy anyway — it was an unresolved bet. A claim triggered by AI output under a silent policy would have gone to coverage litigation with an unpredictable result. The market's move to explicit wording at least makes the question answerable before a claim, which is precisely why the pre-renewal checklist later in this guide exists.

Will My Policy Renewal Add an AI Exclusion in 2026?

More likely than not, based on the direction of filings. Insurance Journal reported that filings with state regulators to use the three ISO endorsements are increasing, and attorney McMullin expects rapid adoption across most lines of business. The change arrives as an endorsement page inside your renewal documents — not as a phone call or a highlighted summary. Some carriers will instead keep pricing AI risk into the premium, so the outcome varies by insurer, but assuming your wording is unchanged is the one assumption the evidence does not support.

The practical response is mechanical, not anxious. Request the complete endorsement set from your broker 60 to 90 days before renewal, compare it page by page against your current wording, and ask specifically where the carrier stands on AI-related claims. A broker who cannot answer has not asked — make them ask, in writing.

Who Pays if an AI Tool You Use Harms a Client: You or the Vendor?

In practice, you do — first, at least. A claim gets filed against the business whose name is on the product, the website, or the advice; recovering from an AI vendor afterwards depends entirely on their terms of service, their pockets, and years of litigation. Your own liability policy is the front line, which is exactly why the exclusions above matter. And the biggest test of where AI liability ultimately sits is now working through the courts.

The Government of British Columbia sued OpenAI in California — reported on September 21-22, 2026 — over the school shooting in Tumbler Ridge, British Columbia, on February 10, 2026. The shooter, Jesse Van Rootselaar, killed five students, a teacher, the teacher's mother, her half-brother, and herself, and had used ChatGPT. The province's allegations are stark: that OpenAI had internally flagged and banned one of her accounts in 2025 but did not warn police, that a second account went undetected, that the company "designed a dangerous product," "ignored the warnings of their own safety team," and was "refusing to notify authorities." The filing states: "One telephone call to the RCMP could have prevented the tragedy."

British Columbia's Attorney General, Niki Sharma, is seeking a jury trial, compensation, and injunctive relief, in a suit described as "separate and parallel" to the victims' families' cases filed in April 2026. OpenAI reportedly refused to disclose the chats, issued a statement on the tragedy and its "ongoing safety work," told Canadian lawmakers in February that updated policies would now refer such users to police, and saw Sam Altman publish an open letter to Tumbler Ridge in April. Trade press has already framed the case as a test of AI liability coverage.

Why should a small-business owner care about a government suing a lab in California? Because the same causal logic — an AI output contributed to harm, and the question is who bears it — will one day arrive at a much smaller doorstep: your chatbot, your AI-drafted engineering note, your generated marketing claim. When it does, the first document examined will be your policy, and the first page turned to will be the exclusions. If you want the wider picture of how non-injury losses route through different policies, read our guides on cyber insurance for small businesses in 2026 and business interruption insurance.

Do AI Exclusions Apply to Professional Liability and E&O Too?

Yes — and in some corners the management-liability market moved earlier and harder than general liability. Insurance Journal reported that insurer Berkley introduced an "absolute AI exclusion" the year before, across directors-and-officers, errors-and-omissions, and fiduciary policies. Its reach is remarkably broad, touching the use or development of AI, AI-generated content, the failure to identify third-party AI content, a company's AI policies and procedures, and the breach of AI-related duties.

Read that list again from an owner's chair. It covers not only AI you build, but AI your staff use and AI content produced by third parties that your business fails to catch. For consultancies, agencies, and any firm whose deliverable is words, code, or advice, professional liability is precisely where an AI claim would land. If your E&O renewal arrives with new AI wording, the stakes are the same as for general liability — only the coverage line differs.

Can I Buy Affirmative AI Coverage as a Small Business?

Yes, though the market is young and uneven. The options, in rough order of cost, are: a buy-back endorsement that reinstates the cover the ISO exclusions removed; a dedicated affirmative AI policy; or a technology errors-and-omissions policy whose wording contemplates AI outputs. The insurtech Counterpart launched affirmative AI coverage in late 2025, according to Yahoo Finance reporting, establishing that the buy-back track exists for buyers who push for it.

Expect two realities when you ask. First, underwriting will want to know exactly how you use AI — which tools, which workflows, what human review — so document that before the call. Second, capacity and pricing vary widely between carriers because there is not yet a settled claims history to price from. Ask your broker for the carrier's written AI position, and treat "we don't have one" as an answer about the broker, not the market.

What Should I Check in My Policy Wording Before Renewal?

Check the endorsement pages, not the declarations — exclusions live in the forms, and the ten checks below take about an hour with your current policy in hand. Do it 60 to 90 days before renewal, while you still have leverage to negotiate, re-quote, or restructure contracts with your AI vendors.

Request the full endorsement set early. Sixty to ninety days out, before auto-renewal locks anything in.
Diff the wording. Compare the new forms line by line against last year's; ask your broker to mark up the changes.
Search for the trigger words. "Artificial intelligence," "generative," "machine learning," "algorithm" — in exclusions, definitions, and conditions.
Map your AI touchpoints to coverage lines. Chatbot advice, AI-drafted content, generated code, AI-assisted hiring, AI in your product — list which policy each one would claim under.
Check E&O and D&O wording, not just general liability. The Berkley-style absolute exclusions live in management liability.
Ask for the carrier's AI position in writing. Not the broker's guess — the carrier's letter or underwriting guidance.
Price a buy-back. Get a quote for reinstating AI cover via endorsement, so the decision is a number, not a fear.
Shift risk contractually. Add indemnification clauses to your AI vendor agreements so their terms, not your policy, absorb their failures.
Document human oversight. A written review process for AI outputs both improves your underwriting story and matters if a claim ever tests your wording.
Re-quote annually. The AI coverage market is repricing yearly; loyalty has no premium discount here.

Source: Lowenstein Sandler LLP — official YouTube channel

How Much Does AI Liability Insurance Cost for a Small Business?

There is no public, standardized price schedule — and any specific figure you see quoted online for "AI liability insurance for small business" should be treated with caution, because premiums are set case by case. Underwriters price on your industry, revenue, limits and deductibles, how deeply AI is embedded in your operations, and the quality of your human-oversight controls. A buy-back endorsement typically adds premium to your existing policy; a standalone affirmative policy is quoted fresh.

The defensible way to budget is comparative, not absolute. Ask two independent brokers to quote the same specification — your limits, your AI inventory, your oversight documentation — and compare the spread. If the two quotes diverge wildly, that gap itself is the market telling you the risk is not yet settled, and it is a strong argument for the contractual risk-transfer steps in the checklist above while the wording market matures.

Are Insurers Required to Notify Me When They Add an AI Exclusion?

In practice, the notice you get is the renewal documentation itself — the endorsement pages arrive with your renewal pack, and the legal onus to read them sits with the policyholder. The forms themselves pass through state regulator approval before carriers use them, as Insurance Journal's reporting on the filings boom describes, but regulatory approval of a form is not a personal warning to you. Nobody will call.

So build your own tripwire. Calendar your renewal date ninety days ahead, request the full new wording every year without being asked, and require your broker to confirm any AI-related changes in writing. The businesses that get caught by exclusions are rarely the ones that read their forms and disagreed — they are the ones whose card on file renewed a policy they never opened.

The Question Nobody Answers Yet

Here is the question every broker, insurer, and attorney in this market is quietly waiting on: when a court finally rules on whether one of these exclusions bars a specific AI claim, what will the answer be? Attorney McMullin's assessment, per Insurance Journal, is that no bellwether AI-exclusion coverage case exists yet. The forms are being attached to policies faster than any judge has interpreted them, and the British Columbia suit against OpenAI — whatever it decides about the vendor — is only the first chapter of a much longer coverage story.

My answer, for what a small-business owner should do while the courts catch up: assume AI-related claims are uninsured until your own broker confirms otherwise in writing, and spend the intervening hour on the ten-point checklist above. Between rising software bills and the health-insurance cost curve facing small employers into 2027, the businesses that win the next few years will be the ones that read the fine print before it reads them. For more practical, data-first guides on running and protecting a lean business, visit truescho.com.