Insurance is one of those things a small business sets up, pays for, and mostly forgets about, trusting that if something goes wrong the coverage will be there. That trust is exactly why a quiet change in 2026 deserves your attention, because it may have opened a hole in coverage you assume you have without your ever being told. As businesses have woven AI into their work, insurers have grown wary of the new and hard-to-predict risks it creates, and they have responded by giving themselves a way to exclude AI-related claims from standard policies.
This is not a reason to panic, and it is certainly not a reason to stop using AI, which remains enormously valuable. But it is a reason to do a specific piece of homework that most small businesses have not thought to do, because an insurance gap is precisely the kind of risk that stays invisible right up until the moment it matters, and by then it is too late to fix. This article explains what changed in plain terms, who is most exposed, what kinds of AI-related problems could leave you uninsured, and the straightforward steps to make sure you are actually covered for the way your business now works.
In January 2026, new insurance endorsements gave carriers a standardised way to exclude AI-related claims from the general liability and professional policies small businesses rely on, creating a coverage gap: if AI contributes to a costly mistake, you may not be insured for it. The businesses most exposed are those using AI in client-facing work, such as marketing agencies, consultants, and professional services, where an AI error can directly harm a client and trigger a claim. The fix is simple homework: ask your insurer or broker directly whether your policies now exclude AI-related claims, understand where your gaps are, and close them through the affirmative AI coverage and standalone AI liability policies that now exist. Do not stop using AI, just make sure your insurance matches how you actually work.
What actually changed
The concrete change is that in January 2026, the body that develops standardised insurance policy language introduced new endorsements, essentially standard clauses that carriers can attach to policies, which allow insurers to remove coverage for AI-related claims from the general liability policies small businesses have depended on for decades. Before this, coverage for AI-related problems was often ambiguous, neither clearly included nor clearly excluded, a situation the industry came to call silent AI. The new endorsements end that silence, but they end it in the direction of exclusion, giving carriers a clean way to say AI-related claims are not covered.
The reason insurers did this is understandable from their side. AI introduces new kinds of risk that are hard to quantify with the historical data insurance pricing relies on, and rather than take on an open-ended exposure they do not yet know how to price, many carriers would prefer to exclude it and offer coverage for it separately and deliberately instead. That is a rational move for an insurer, but it shifts a risk onto the business, because a policy that used to cover you broadly may now carry a carve-out for exactly the kind of AI-involved mistake that is becoming more common as more businesses use AI.
The crucial and uncomfortable part is that this change can happen without you noticing. Exclusions get added at renewal, buried in policy language that few small business owners read closely, and the first time many businesses will learn that their AI-related coverage was removed is when they file a claim and it is denied. That is the worst possible moment to discover a gap, which is exactly why the sensible response is to find out now, proactively, rather than assuming the coverage you have always had is still the coverage you have.
The coverage gap explained
To see why this matters, picture how an AI-related claim actually arises. Your business uses an AI tool to help with some piece of work, the AI produces something wrong, inaccurate information, a flawed recommendation, a mistaken output, and that error causes harm to a customer or third party who then seeks compensation. In the past, your general liability or professional liability policy might well have responded to that claim, because it was simply a mistake made in the course of your business, and mistakes are what such policies exist to cover. The new exclusions can change that answer to no.
The gap, then, is the space between how your business now operates, using AI in various tasks, and what your insurance now covers, potentially everything except the AI-involved mistakes. As AI becomes more woven into daily work, more of your potential mistakes have an AI element somewhere in their history, which means the excluded category is not some rare edge case but a growing share of the ways things could plausibly go wrong. A gap that might have been trivial when nobody used AI becomes significant precisely because AI is now everywhere in how work gets done.
It is worth naming that this is genuinely a new category of business risk, distinct from the more familiar AI risks like the vendor-availability problem we wrote about in our piece on the Anthropic export ban. That earlier risk was about a tool disappearing. This one is about being financially exposed when a tool you use contributes to a mistake, and your insurance declining to cover the fallout. Both are manageable, but both require you to look at a corner of your business, in this case your policies, that you probably have not examined through an AI lens before.
Who is most exposed
Not every business faces this risk equally, and it is worth being specific about where the exposure concentrates so you can gauge your own. The most exposed businesses are those that use AI in client-facing or deliverable-producing work, where an AI error can flow directly into something a customer relies on and then cause them harm. Marketing and creative agencies are a clear example, because they lean heavily on AI for content, copy, images, and social media, and a mistake in that output can directly damage a client. Professional services firms, consultants, financial planners, and similar are also significantly exposed, because they use AI to help draft client deliverables and reports, and when the AI gets something wrong the resulting harm often shows up as a professional liability claim.
The common thread is that these businesses use AI in the production of things other people depend on, so an AI error does not just create an internal problem, it creates a potential liability to a third party who was relied on the flawed work. That is precisely the situation where you would expect your professional or general liability insurance to respond, and precisely the situation the new exclusions may carve out, which is why these client-facing, deliverable-producing businesses have the most reason to check their coverage carefully and promptly.
Businesses that use AI more internally, to automate their own admin, handle their own data, or streamline their own operations without producing AI-influenced work that customers directly rely on, face a smaller version of this specific risk, though it is still worth understanding their coverage. The point is not that every business is in acute danger, but that the exposure tracks how directly your AI use touches other people, and any business whose AI-assisted work reaches customers or third parties should treat the coverage question as genuinely important rather than theoretical.
What kinds of claims are at risk
The category of AI-related problem that most concerns insurers, and that appears most often in discussions of this risk, is the AI mistake that produces harm through wrong information. Surveys of businesses have found that AI errors, misinformation, and so-called hallucinations, where an AI confidently generates inaccurate information, rank as the leading AI-related risk concern overall. When that inaccurate output makes its way into something public-facing or into work a client relies on, it can cause real harm, damage to reputation, financial loss, or worse, and harm to a third party is what turns an internal mistake into a liability claim.
Concretely, the kinds of claims at risk include situations where AI-generated content contained an error that misled or damaged someone, where an AI-assisted recommendation or analysis turned out to be wrong and a client acted on it to their detriment, or where AI-produced marketing or communications caused some form of harm or legal exposure. These are not exotic hypotheticals, they are ordinary ways that ordinary business mistakes happen, now with an AI element in the chain of causation, and it is exactly that AI element that the new exclusions may seize on to deny coverage.
This connects directly to the discipline we emphasised in our piece on AI copyright and content: keeping a human meaningfully in control of AI output is not only good for quality and ownership, it is part of managing your liability. A human reviewing and taking responsibility for AI-assisted work reduces the chance that an AI error reaches a customer uncorrected, which reduces the chance of a claim in the first place. Good practice in how you use AI and proper insurance coverage for when something slips through are complementary defences, and a well-run business wants both.
What to actually do
The first and most important step is simply to ask, directly and specifically. Contact your insurer or broker and ask in plain terms whether your current policies now exclude AI-related claims, and if so, exactly what that exclusion covers. Do not assume, and do not rely on the coverage you had a year ago, because the whole point of this change is that exclusions can be added quietly at renewal. A direct question gets you a direct answer, and knowing where you actually stand is the entire foundation of managing this risk, since you cannot close a gap you have not confirmed exists.
Once you understand your position, close any gap you find, which is very achievable because the insurance market has responded to this problem with solutions. Affirmative AI coverage is now available through what are called write-back endorsements, which add AI coverage back onto a policy explicitly, and through standalone AI liability policies designed specifically for the AI-related risks small and medium businesses face, covering losses that standard general liability policies now exclude. The gap the exclusions created is matched by products built to fill it, so the practical task is to identify the right one for how your business uses AI and put it in place.
Finally, treat this as part of a periodic review of how your insurance matches how your business actually operates, because the AI coverage question is really an instance of a broader discipline. Businesses evolve, adopt new tools, and take on new kinds of work, and coverage set up for an earlier version of the business can quietly fall out of step with the current one. A short, regular check that your policies reflect what you actually do now, with AI use explicitly on the checklist, keeps you from carrying invisible gaps of any kind. Understanding your real exposures across the business is exactly the kind of clear-eyed review our €49 audit is built to support.
The bottom line
A genuine and largely unnoticed change happened to small business insurance in 2026: new endorsements gave carriers a standardised way to exclude AI-related claims from the standard policies businesses have relied on for decades, opening a coverage gap exactly as AI becomes woven into how work gets done. The danger of this gap is its invisibility, because exclusions get added quietly at renewal and many businesses will only discover the hole when a claim is denied, which is the worst possible moment to find out.
The response, though, is neither fear nor retreat from AI, which remains far too valuable to give up over an insurance technicality. It is straightforward homework: ask your insurer or broker directly whether your policies now exclude AI-related claims, understand where your gaps are, and close them through the affirmative AI coverage and standalone policies that the market has created for exactly this purpose. Pair that with the good practice of keeping a human in control of AI-assisted work, so fewer errors reach customers in the first place, and you have both fewer claims and proper coverage for the ones that happen. Keep using AI, and make sure your insurance simply catches up to the way your business already works.
Sources
- Business Insurance USA — The AI Coverage Gap Hitting Small Business Insurance Renewals in 2026
- Lathrop GPM — The AI Coverage Gap: What New Insurance Exclusions Mean for Your Business
- Insurance Business — AI liability emerges as "the new cyber" for SMEs
- Fenwick — The End of "Silent AI"? Emerging AI Exclusions and Coverage Fragmentation
- Munich Re / HSB — Introducing AI Liability Insurance for Small Businesses
- Ward and Smith — The Insurance Implications of AI in Your Business
- NEXT Insurance — AI and the Future of Small Business Insurance: 2026 Predictions
- Policyholder Pulse — AI Exclusions in Insurance Policies: Broad Language, Uncertain Impact