The renewal pack lands and the proposal form carries a question that wasn’t there last year. Does your business use artificial intelligence in delivering its services? Plenty of owners tick yes without a second thought, because for a growing share of firms the honest answer is yes. What has changed is the wording sitting behind the question. For years, AI risk was absorbed silently into professional indemnity and cyber policies, treated as just another piece of software. Insurers are now redrawing those boundaries, and the redrawing happens at renewal, in endorsements few people read.
What is AI insurance cover?
AI insurance cover usually means a set of existing policies rather than one product. For an owner-managed services firm, it is the combination of professional indemnity, cyber and management liability cover, plus any AI-specific endorsements or standalone add-ons, that responds when your use of AI causes harm to a client, a customer or a third party. The load-bearing question is which of those wordings still respond.
The market is splitting in two directions at once. US law firm Fenwick describes the era of “silent AI”, where policies covered AI incidents simply because nothing excluded them, as ending. Standard-form endorsements now let insurers exclude losses “arising out of” generative AI, a definition broad enough to catch text, images, code and automated responses. Some carriers have gone further, and analysis by Lathrop GPM found absolute exclusions in directors’ and management liability policies that disclaim any claim attributable to a company’s use of AI, whether the tool was built in-house or bought off the shelf.
The other direction is affirmative cover. In March 2026, specialist insurer HSB, part of Munich Re, launched AI liability insurance aimed at smaller firms, covering bodily injury, property damage and personal and advertising injury arising from AI use. Its example scenarios are strikingly ordinary, an AI-controlled heating system causing a slippery floor, AI-generated installation instructions causing a leak. Cover is being redefined around AI in both directions, and neither direction announces itself.
Why does it matter for your business?
Because AI is already inside the work your insurance protects. A 2024 YouGov poll of leaders at smaller UK businesses found 31 per cent already using AI tools and another 15 per cent planning to, with task automation and marketing content the commonest uses and one in five using AI for decisions. When those outputs go wrong, the claim lands on your policy, whatever the tool was.
The exposure sits close to revenue. Marketing content drafted with AI can defame a competitor or infringe copyright, and a chatbot can misstate what your service does. An AI-assisted report with a wrong number in it is, in legal terms, a professional error like any other, and a client who relies on it will claim against you regardless of which tool produced the mistake.
Regulated firms show where this is heading. The FCA reports that around three quarters of regulated financial firms have adopted some form of AI, and that those firms rank cybersecurity as the biggest risk it brings. Deepfake voice fraud and AI-crafted phishing are already in the threat toolkit, which is why AI-aware cyber cover matters as much as professional indemnity. Good cover treats AI-assisted professional errors, AI-generated content claims and AI-enabled fraud the way it treats their human-made equivalents.
Where will you actually meet it?
In four places. The renewal proposal form, which increasingly asks direct questions about AI use. The endorsements attached to your policy, where phrases such as “arising out of generative artificial intelligence” carve out cover. Your broker conversation, which you will probably have to steer yourself. And a small but growing shelf of specialist products built for AI risk.
The proposal form is the easy one to spot. The endorsement is harder, because it arrives as a page of definitions inside the renewal documents, and the phrases to search for are “artificial intelligence”, “automated decision-making”, “algorithmic outputs” and “generative AI”. A wording that was silent last year may now carve out precisely the claims your working practices generate.
The specialist shelf is worth knowing even if you never buy from it. HSB’s product writes affirmative AI liability cover for smaller firms. Munich Re’s aiSure does something different again, guaranteeing an AI model’s agreed performance thresholds, which matters if you ever promise a client that a model will hit a specific accuracy or conversion figure. Standard professional indemnity covers negligent failure, and a voluntary performance promise sits outside it. If your contracts contain hard AI performance guarantees, that exposure is yours until you close it.
When should you ask for explicit AI cover, and when can you leave it?
Ask when AI outputs reach clients or shape decisions about people and money. If AI drafts advice, screens candidates, prices work or talks to customers, get written confirmation of how each policy treats it. Leave it, for now, when AI is supervised internal drafting, your wordings carry no AI exclusions, and a qualified human reviews everything before it leaves the building.
The useful test is the junior employee. A junior’s mistakes are covered because a qualified person supervises the work, and insurers are settling on the same logic for AI. Broker PIB, summarising regulator guidance for law firms, recommends documenting AI inputs and outputs, keeping human review on anything that reaches a client, and training staff on the tools’ limits. Meet that standard, keep wordings free of AI exclusions, and existing professional indemnity plus cyber will usually carry the load.
Know what no policy will pick up. Reckless, unsupervised use sits outside cover, and the courts are already producing cautionary examples of professionals sanctioned over AI-fabricated citations. Losses from believing vendor hype stay on your balance sheet too. In 2025 the US Federal Trade Commission sued Air AI, which had promised small firms that its conversational AI could replace customer service teams and generate extraordinary earnings. Buyers who lost money had made a bad purchase, and a bad purchase is not an insurable event. Regulatory fines rarely transfer either. The ICO can fine up to £17.5 million or 4 per cent of worldwide turnover, and UK policies will fund your lawyers and incident response long before they pay the penalty itself.
Which related terms should you know?
Five terms carry the weight in these conversations. Silent AI cover, professional indemnity, technology errors and omissions, performance guarantees and high-risk AI classification. Knowing what each one means lets you read an endorsement, brief your broker and spot the difference between a policy that responds to AI-assisted work and one that has been rewritten around it.
Silent AI cover is the implicit protection you hold when a wording says nothing about AI, and it is the thing currently being written out of the market. Professional indemnity and technology errors and omissions are the covers that respond to negligent professional work, AI-assisted or otherwise. Performance guarantees, aiSure being the visible example, insure a promise about model performance rather than a negligent failure, and the two are priced and worded very differently. High-risk AI classification comes from the EU AI Act, which reaches firms whose AI outputs are used in the EU and treats employment screening and credit decisions as high-risk categories with heavier duties attached.
The final term is really a principle. The National Cyber Security Centre’s guidance on cyber insurance makes the point that a policy complements security controls rather than replacing them, and the same holds for AI cover and AI governance. An insurer will expect supervision, documentation and sane procurement before it pays a claim. The businesses that get full value from AI insurance are the ones that needed to rely on it least.
Where does that leave the renewal question on the form? Answer it accurately, then turn it around. Ask your broker, in writing, which policy responds if AI-assisted work causes a client loss, whether any AI endorsements have been added this year, and how AI-enabled fraud is treated under your cyber cover. An hour spent mapping where AI touches your client-facing work, and one direct letter to your broker, buys more protection this quarter than any new product on the shelf.
