FCA authorised · FRN 724952 0117 325 0027Quote & buy →
AI insurance
AI insurance is insurance that expressly covers the liabilities a business faces from building, selling or using artificial intelligence, instead of leaving AI risk to general wording. It matters because a policy that never mentions AI may be unclear or restrictive when an AI-related claim arrives.
Part of: AI liability guides
In short
AI insurance usually combines professional indemnity, cyber, product or public liability and IP infringement cover, arranged so the policy names your AI work as insured activity. Firms that build or sell AI, AI consultancies and integrators, and professional firms using AI in client work all carry AI exposures. Under the Insurance Act 2015 a business must make a fair presentation of the risk, so describing your AI work precisely protects the cover as well as defining it. US exposure changes the risk, must be declared, and cover for it can be arranged subject to underwriting.
Most business insurance wordings were drafted before AI was part of everyday work. Whether an AI-related loss falls inside a PI, cyber or liability policy depends on how the policy describes what you do and whether it contains AI exclusions or clarifications.
AI insurance makes the cover affirmative: the policy says, in terms, that your AI work is insured activity. A growing number of policies in the market now give affirmative AI cover, and approaches vary. It usually answers four questions:
For a short definition, see our AI insurance wiki entry. For the difference between cover that names AI and cover that says nothing about it, see affirmative AI cover vs silent AI.
AI exposure is not confined to AI companies. Most of it sits with three groups of firms.
Model developers, software businesses with AI features, agent platforms, and makers of AI-enabled devices or robotics. The exposure sits in the product: wrong outputs, autonomous actions that go wrong, data handling, and allegations that the model or its outputs infringe third-party rights. The question often arises when an enterprise customer or investor asks for evidence of cover; see insurance for AI startups.
Consultants, agencies and integrators who select, configure, fine-tune or connect AI tools for clients. The client relies on your judgement, so a poor recommendation or faulty integration is a classic professional indemnity exposure. More on PI for AI consultants.
Accountants, lawyers, recruiters, marketers and other professionals who use AI to draft, research or analyse. Your duty of care to the client does not change because a tool produced the draft. Their PI needs to respond clearly to AI-assisted work; see PI insurance for firms using AI. If you run agents that send emails, make payments or change code, read AI agents insurance.
The scenarios below are illustrative, not real cases. Whether a policy responds always depends on its terms, its exclusions and the facts.
| Cover | What it usually responds to | Illustrative AI claim |
|---|---|---|
| Professional indemnity | Claims that your advice, service or technology caused a client financial loss through negligence, error or omission | An integrator configures an AI screening tool for a recruitment firm that filters out qualified candidates for months. The client claims lost fees from the integrator. |
| Cyber | Data breaches and network security failures, with related costs such as investigation, notification and third-party claims | Crafted prompts trick a customer service chatbot into revealing other customers' order details, causing incident costs and data protection complaints. |
| Product / public liability | Third-party bodily injury or property damage caused by your products or your business activities | A warehouse robot's AI vision misjudges a pallet edge and the robot strikes a visiting contractor, who claims against the robot's supplier. |
| IP infringement | Allegations that you infringed someone else's copyright, trade mark, database right or similar rights | An agency's AI image tool produces campaign artwork closely resembling a photographer's published work. The photographer alleges copyright infringement. |
One AI incident can touch more than one of these. Holding the options in one policy can reduce argument over which policy should respond. Read more on each: does PI cover AI mistakes, cyber insurance and AI risks, AI product and public liability insurance and AI IP infringement insurance.
PI and liability policies define cover by reference to the business described in the schedule. If it says "IT consultancy" and a claim arises from a model you trained for a client, the insurer may question whether that work was insured. Writing the AI work into the policy settles the point at the start, not at claim time.
For business insurance, section 3 of the Insurance Act 2015 requires the insured to make a fair presentation of the risk before the contract is entered into. That means disclosing every material circumstance the insured knows or ought to know or, failing that, enough to put a prudent insurer on notice that it needs to ask more (section 3(4)). Under section 7(3), a circumstance is material if it would influence a prudent insurer's judgement on whether to take the risk and on what terms, and the Act's examples include special or unusual facts relating to the risk. The duty also applies when a policy is varied (section 2(2)).
If the insurer shows that, but for a breach, it would not have written the policy or would have done so on different terms, it has a remedy (section 8). Under Schedule 1, remedies range from avoiding the policy and keeping the premium for a deliberate or reckless breach, to applying the different terms, or paying a proportion of a claim where a higher premium would have been charged. AI work that is new, high-volume or customer-facing is the kind of fact a prudent insurer would want to know. See our guide to the duty of fair presentation.
"Silent AI" describes AI exposure sitting in a policy that neither expressly includes nor excludes it. The idea comes from silent cyber: the PRA's supervisory statement SS4/17 describes non-affirmative cyber risk as exposure in policies that do not explicitly include or exclude cyber, and expects insurers to manage it, for example by offering explicit cover, introducing exclusions or attaching specific limits. Commentators report a similar shift for AI, with some insurers adding AI exclusions or narrowing definitions. Silent cover may or may not respond; affirmative cover tells you where you stand. See AI exclusions and clauses in insurance policies and silent cyber and affirmative cover.
Most AI liability claims are brought under existing law: negligence, contract, misrepresentation, data protection, product liability and IP.
| Development | Position at October 2026 | Why it matters |
|---|---|---|
| UK AI regulation | No general AI statute. The government's February 2024 response to its AI white paper asks existing regulators to apply five cross-sector principles. As at September 2026 no government AI bill had been introduced (Lewis Silkin). | Contracts and professional duties remain the main sources of liability. |
| FCA | On 27 January 2026 the FCA said it does not plan to introduce AI-specific regulation and will rely on its existing framework, including the Consumer Duty. | Regulated clients expect AI suppliers to support their compliance, often by contract. |
| UK data protection | ICO guidance applies data protection law to AI systems that process personal data (under review after the Data (Use and Access) Act). | Personal data in training sets or prompts creates cyber and PI exposure. |
| EU AI Act, Regulation (EU) 2024/1689 | Article 2 reaches non-EU providers placing AI systems on the EU market, and non-EU providers and deployers whose output is used in the EU. After the Digital Omnibus on AI, Article 113 applies most high-risk obligations from 2 December 2027 (Annex III) and 2 August 2028 (Annex I). | UK firms selling into the EU may take on duties, and customers may pass them down by contract. |
| EU Product Liability Directive (EU) 2024/2853 | Treats software, including AI systems, as a product. Applies to products placed on the EU market after 9 December 2026. | Defective AI products sold into the EU carry product liability exposure. |
| Moffatt v Air Canada, 2024 BCCRT 149 | A Canadian tribunal (14 February 2024) held an airline liable in negligent misrepresentation for its website chatbot's wrong answer on bereavement fares. | A business can be held to what its AI tells customers. Not binding in UK courts. |
| Getty Images v Stability AI [2025] EWHC 2863 (Ch) | The High Court (4 November 2025) held the trained model was not an infringing copy and rejected the secondary copyright claim; trade mark findings were limited. Training took place abroad, so its lawfulness in the UK was not decided. Getty was given permission to appeal. | IP risk from AI training and output remains unsettled in the UK. |
Selling to US customers can change an AI risk significantly. Claims are defended under US procedure and costs, US federal courts can certify class actions under Rule 23 of the Federal Rules of Civil Procedure (Federal Judicial Center guide), and states are legislating on AI. Colorado, for example, has repealed and re-enacted its 2024 AI law through SB 26-189, signed in May 2026, with disclosure duties for automated decision-making technology from 1 January 2027.
That is why insurers ask for your USA and Canada turnover, and why territorial and jurisdiction clauses matter as much as the limit. Some policies cover worldwide activities but exclude claims brought in US or Canadian courts. Through Apex's AI scheme, cover for US exposure is available and can be arranged subject to the insurer's underwriting and the policy terms. See AI insurance with US cover for UK firms.
Apex has an AI insurance scheme with an insurer in which the policy names your specific AI work as insured activity, rather than leaving it to a generic business description. The cover options are professional indemnity, cyber, product or public liability, and IP infringement, so several AI exposures can sit in one policy and you choose the options you need. Cover for US exposure is available, subject to the insurer's underwriting, and all cover is subject to the policy terms.
To quote, we need:
You can start an AI insurance quote online. The form takes a few minutes.
AI insurance is cover that expressly includes the liabilities a business faces from building, selling or using artificial intelligence. It usually combines professional indemnity, cyber, product or public liability and IP infringement cover, with the AI work named in the policy so there is less room for dispute about whether it was insured. Cover is always subject to the policy terms.
Not always a separate policy, but your existing cover needs to respond to AI-assisted work. Check how your PI policy describes your services, look for AI exclusions, and tell your insurer how AI is used in client work. A broker can confirm whether affirmative AI cover would be clearer.
Silent AI is AI exposure in a policy that neither expressly covers nor excludes it. The term borrows from silent cyber, which the PRA describes as cyber risk in policies that do not explicitly include or exclude it. Silent cover may or may not respond to an AI claim, which is why affirmative wording is increasingly sought.
Yes, cover for US exposure is available, subject to the insurer's underwriting and the policy terms. Insurers will ask how much turnover comes from the USA and Canada, and will look closely at jurisdiction clauses and the contracts you sign with US customers. Declaring this accurately is part of a fair presentation.
Apex needs a precise description of what your business does and its AI work, your address, your turnover including how much comes from the USA and Canada, and the limit of indemnity you want. The online form takes a few minutes.
It can, where IP infringement cover is included and the claim falls within its terms. IP sections commonly exclude deliberate infringement and some exclude patents or liability taken on by contract. UK law on AI training and copyright is still developing, so check the wording carefully and describe how you use training data and outputs.
Tell us what your AI work is, where you are based, your turnover and the limit you need. It takes a few minutes online, and a broker will come back to you with options. Or call 0117 325 0027.
Start an AI insurance quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information, not advice on your individual circumstances, and it does not guarantee that cover will be available or on what terms.
