FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: IT consultants and developers
IT contracts should now say whether and how AI is used, what happens to client data, what is warranted about AI outputs and how liability is capped. Those terms decide much of your exposure when an AI-related project goes wrong, and whether your PI will match it.
Part of: AI in IT consultants and developers
In short
A statement of work for AI-assisted or AI-built delivery should cover disclosure of AI use, restrictions on client data (including bans on using it to train AI, as PPN 017 allows public buyers to impose), roles under the AI Cyber Security Code, testing and acceptance, warranties for AI outputs, IP indemnities and a liability cap. There is no statutory or BCS PI requirement for IT consultants, so contracts set the cover clients expect. Technology PI usually responds to liability you would have in law anyway, subject to the wording; promises that go further can create uninsured exposure.
Traditional IT contracts assume a person writes the code, reviews the documents and makes the decisions. AI-assisted delivery muddies that: data may go to third-party model providers, outputs are probabilistic, and a delivered AI system may behave differently over time. Clients are responding with new demands, and public sector buyers have formal guidance. Without clear terms, disputes about what was promised are more likely, and you may be held to expectations you never agreed.
Under PPN 017, applying to procurements from 24 February 2025, public buyers may ask suppliers to declare AI used in bids and in service delivery, and may verify AI-assisted content. Private clients increasingly ask similar questions. A short clause describing where AI tools assist delivery, and the human review applied, sets expectations and supports the honesty required by BCS clause 3.5 on performance information. Avoid broad statements such as "no AI is used" unless you can guarantee them across the team.
PPN 017 also allows contracts to bar the use of buyer data to train AI without written approval. Many private clients now want the same. Your contract should say which AI tools may receive client data, whether providers may retain it or use it for training, where it may be processed, and how it is deleted. Make sure your own tool settings and suppliers can actually comply. For BCS members, clause 3.4 on confidentiality applies regardless. See LLM integrations and data leaks and training models on client data.
The AI Cyber Security Code of Practice distinguishes developers, who create or adapt AI systems, from system operators, who deploy them in their infrastructure. Your statement of work should say which role you take, who monitors the system after go-live, who updates it and who decides what an AI system may do autonomously. Testing scope, including adversarial testing for chatbots and agents, and acceptance criteria should be explicit. See building AI chatbots and deploying AI agents.
Clients may ask for warranties that an AI system is accurate, unbiased, secure or legally compliant. Generative AI cannot guarantee accuracy, so absolute warranties are risky. A safer approach is to warrant the process: that you will use reasonable skill and care, follow agreed security practices, and build specified safeguards, such as the Article 22C features for automated decision systems. Fitness for purpose wording can impose strict liability that goes beyond negligence.
A liability cap should be set at a level your PI limit supports, with clear exclusions for indirect loss where appropriate. Watch for carve-outs that make data protection or IP liability uncapped, and for indemnities covering "any loss caused by AI outputs". Contractual indemnities for IP infringement are common in AI work; check how they interact with your cover. Our pages on AI contract clauses and indemnities and limitation of liability in IT contracts go further.
Technology PI is generally written to cover your legal liability for negligent professional services. Many wordings limit or exclude liability you take on only because of a contract, such as uncapped indemnities, penalties, guarantees or fitness for purpose warranties, unless that liability would have existed anyway. So the more your contract promises beyond reasonable skill and care, the larger the gap may be between what you owe and what your policy pays. A broker can compare draft terms with your wording before you sign. See does tech PI cover AI failures?.
Apex arranges technology PI for IT consultants and developers. A broker can review AI clauses in a draft contract against your policy and flag terms that may sit outside cover. See the IT consultants AI hub.
There is no general legal rule requiring it, but contracts increasingly do. Public buyers using PPN 017 may ask suppliers to declare AI used in bids and service delivery. Being accurate about AI use also helps meet BCS clause 3.5 on not misrepresenting performance.
Yes. PPN 017 expressly allows public sector contracts to bar the use of buyer data to train AI without written approval, and private clients often include similar terms. Check that the AI tools and providers you use can comply.
Usually not in absolute terms. Generative AI cannot guarantee accuracy. Warranting reasonable skill and care, agreed testing and specified safeguards is safer and more likely to align with what tech PI covers.
Set it at a level your PI limit supports and avoid uncapped carve-outs for AI outputs, data or IP unless you have checked your cover. Discuss significant contracts with your broker before signing.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI 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 about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.
