AI and professional negligence: the new risk for professionals

A version of this article appeared in the July issue of Engineering Designer Magazine

AI is no longer a futuristic talking point for professionals. It is becoming part of day-to-day practice in every profession, including construction and engineering.

But this creates a new negligence question: if something goes wrong, will the professional be criticised for using AI, or for failing to use it? Addressing the Professional Negligence Bar Association in May 2024, Sir Geoffrey Vos, the Master of the Rolls, described the emerging dilemma by suggesting that professionals would be “damned if they do and damned if they don’t”.

They may face liability for losses caused by reliance on AI, but they may also face liability for choosing not to use AI in their work. As each industry practice evolves, professionals can’t simply ignore or embrace AI. There is potential liability in either case. Professionals should consider how to responsibly use AI as technology and industry standards evolve.

AI is already changing practice

In 2025, Arup’s global survey of the built environment sector found that 36% of architects, engineers and city planners use AI daily, and over 80% use it at least weekly – not only for simple chatbot tasks, but also for advanced work that reshapes how projects are conceived, delivered and managed. The direction of travel is clear: AI is becoming part of the working environment in which all aspects of professional judgement are exercised.

This matters because, as well as technical concerns about using or not using AI, engineers and construction professionals need to consider how this affects their risk of professional negligence, which is assessed by reference to the reasonable skill and care expected of a competent professional in today’s technological environment. As the environment changes, so do the expectations.

Damned if you do – use AI

Many professional now use AI to accelerate routine and technical tasks: researching standards, interrogating large data sets or summarising regulations. Others may already be embracing the technology on a more fundamental level, asking AI to produce bespoke work and advice for clients.

But if a professional relies on AI output that is wrong, incomplete or poorly understood, and that error feeds into the service provided to the client, they may face a professional negligence claim. This is not because AI is involved, but because the professional remains responsible for the work that reaches the client.

All professionals are liable for the way they use tools, including AI, to provide their work or service because of the duty of care that they personally owe to their clients, both contractually and, in many cases, under the law of negligence. AI is only a tool. Responsibility for checking the reasoning, challenging the output and deciding whether it is safe to rely on remains with the engineering or construction professional.

Any erroneous advice or response given by AI is attributed to the professional, who has responsibility for checking its validity and factual accuracy before relying on it or advising clients to rely on it. If an AI model malfunctions and causes a loss, a professional may be able to make a claim against the provider of the AI, but the client who suffered a loss has a claim against the professional.

In short, AI can enhance construction and engineering work, but it does not dilute professional accountability.

Damned if you don’t – use AI

A more difficult question is whether those operating in the industry may face a ‘new’ type of professional negligence claim in the future if they choose not to use AI when providing professional advice or services. That may sound speculative, but the underlying legal principle is not new. It has long been established that a professional’s conduct is judged relative to the accepted practice of others in the same profession. If industry practice changes, the standard against which conduct is measured can change with it.

This principle is known as the ‘Bolam test’ after Bolam v Friern Hospital Management Committee, the 1957 case in which it was applied. The case concerned possible clinical negligence. The court held that a clinician is not negligent if they act in accordance with an accepted practice that is widely considered proper by a responsible body of medical professionals in that area of practice.

The Bolam test has since been applied to cases involving a range of professionals, including construction professionals, valuers and financial advisers. Modern case law including Montgomery v Lanarkshire Health Board [2015] and McCulloch v Forth Valley Health Board [2023] has also broadened the test, considering situations where the standard of practice should reflect not only professional norms but also the reasonable expectations of clients.

So, the question for construction professionals and engineers may become not whether AI must be used, but whether refusing to use AI could eventually be seen as falling below the standard expected in the profession.

This will not happen overnight, and it may not apply across all tasks equally. But in some areas of construction and engineering practice, AI tools may become so well established that choosing not to use them could become harder to defend.

To ensure that work and services are efficient, expeditious and competitively priced, professionals may begin to be expected to use AI to deliver accurate and up-to-date advice, reducing the risk of human error or oversight. Using AI may become the accepted standard of practice.

A professional who then refuses or fails to use widely adopted AI-assisted tools may face a professional negligence claim for falling below this new standard. For example, not using AI where it is demonstrably safer or more efficient, and where this results in poorer risk identification, slower delivery or higher error rates than the profession’s ordinary practice, may be unreasonable and a failure to meet the expected duty of care. It may begin to look less like professional independence and more like avoidable inefficiency or exposure.

Conclusion – professionals need a defensible approach to AI

As professional practices continue to be transformed by AI, the legal framework surrounding professional negligence is also expected to evolve. Engineers and construction professionals should maintain a careful, defensible balance between innovation and caution, monitoring changes to standards, practice and client expectations and making decisions about using AI in light of these factors.