FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Project managers
Project managers delivering AI implementations take on the usual delivery risks plus some new ones, and a failed AI project can leave the PM facing claims that look like IT consultancy claims. Whether a PM's policy responds depends on how the services were described.
Part of: AI in project management
In short
PMs may be asked to run AI implementations: chatbots, document automation, forecasting tools and AI features in client systems. PMI's CPMAI credential sets out a six-phase methodology covering ethics, testing, monitoring and governance, and ISO/IEC 42001 provides a management system benchmark. AI projects fail in particular ways: poor data, unreliable outputs, unmeasured benefits and governance gaps. When they fail, the client may say its PM should have warned it or managed delivery better. A PM policy may not cover advice that strays into technical design or AI selection, which is IT consultancy territory, so the services description matters.
A PM on an AI implementation typically manages scope, vendors, data preparation, testing, governance sign-offs and change management for users. PMI's PMI-CPMAI credential is built on a six-phase CPMAI methodology for AI projects that emphasises ethical outcomes, testing and monitoring for reliable and explainable results, and governance. APM research, including work on integrating AI with agile project management, also explores how delivery approaches adapt.
AI projects differ from conventional IT projects in that success cannot be fully specified up front. A model may work in testing and drift in use, and the quality of the outcome depends heavily on the client's data.
There is no UK rule specific to managing AI projects. The duty is to use reasonable skill and care as a PM, judged against the services you agreed to provide. ISO/IEC 42001:2023 sets requirements for an AI management system and is cited in construction as a governance benchmark; aligning a project's governance with it gives a defensible structure. Where the AI will make significant decisions about individuals, UK GDPR Articles 22A to 22D, in force from 5 February 2026, require safeguards, and a competent PM would usually flag that. RICS-regulated PMs must also follow the RICS AI standard, including section 5 on developing AI, which requires firms building AI to record application, risks, benefits and alternatives before deployment.
Illustrative scenario. A PM is engaged to deliver an AI document-processing tool for a client's claims team. The PM reports the project as on track through build, but user acceptance testing uses clean sample data rather than the client's real documents. After go-live, error rates are high, staff revert to manual processing and the client writes off the investment. It claims against the PM for failing to manage testing and for recommending go-live, and against the vendor separately.
A boundary issue follows. If the PM also advised which AI product to buy, or designed how it would integrate with the client's systems, the client may frame the claim as technical advice rather than project management.
PM PI policies usually cover the project management services declared. Advice on technology selection, system design, data architecture or model configuration is usually the territory of IT consultancy or technology PI, which is written differently and often combined with cyber. If your role on AI projects goes beyond delivery management, it should be declared, and you may need a policy designed for it; see PI for IT project managers. Contract terms matter: limitation of liability, exclusion of consequential loss and clear scope help in either case. The IT consultant view is in AI in IT consultancy, and the consulting view in AI implementation projects for management consultants.
Apex arranges PI for project managers, including those delivering AI and IT programmes, and a broker can check whether your declared services match your role. See APM, PMI and RICS on AI, does PM PI cover AI errors? and the AI in project management hub.
No, it is voluntary. It can show clients you follow a structured method that covers ethics, testing and governance, but it does not change your legal duties or insurance position.
Only if the client can show the PM fell below the standard of a competent PM and that caused its loss. AI failures can stem from data or the product itself, so the vendor and the client's own decisions are often relevant too.
Not necessarily. Product selection and technical design are usually IT consultancy activities. If you give that advice, declare it to insurers and check whether a technology PI policy is more suitable.
That is a decision for your firm or the client. Some PMs may use its structure to set up project governance without certifying, which still helps show a recognised approach was followed.
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.
