If you're developing, funding or investing in construction, here's a question worth asking: how many of the professionals you've appointed are using artificial intelligence (AI) tools as part of their work on your project?

Compliance checking. Structural analysis. Programme sequencing. Cost planning. The answer, almost certainly, is more than you think. The follow-up question (and the one your contracts are almost certainly not answering) is what happens when one of those tools produces an output that turns out to be wrong?

The regulatory framework puts the liability on a person, not a platform

Under the Building Control (Amendment) Regulations 2025, accountability is personal. The Assigned Certifier signs the compliance certificate. The Design Certifier signs off on design. The regulations aren't interested in what tools were used to get there – they are interested in the name on the document. That means every professional on your project relying on AI-generated outputs is carrying liability for those outputs, whether their appointment document acknowledges that or not.

The problem for you as a client isn't just their exposure. It's yours. If a compliance certificate is later challenged because the underlying analysis was AI-assisted and inadequately verified, the dispute doesn't stay with the professional. It reaches into your project programme, your funding arrangements, and your asset value.

The EU AI Act has already changed the landscape

There is a further layer that many clients aren't aware of. The EU AI Act, which is now in force, classifies certain construction-related AI systems as high-risk. Tools used for safety component assessment, structural integrity analysis, and compliance verification against building regulations all fall within that classification. From August 2026, the obligations that attach to high-risk AI systems are live. That includes requirements around transparency, data quality, and critically, meaningful human oversight of AI outputs before they are relied upon.

What this means in practice is that the professionals on your project who are using these tools aren't just operating under general professional duty obligations. They are operating under active EU regulatory requirements. And the person required to exercise meaningful oversight is the named professional, not the platform provider. Your appointment documents need to reflect that reality.

What your appointments are probably not saying

Most professional appointment documents in use on Irish construction projects today were drafted before AI tools became a routine part of practice. That means they are likely silent on:

• Whether the professional is permitted to use AI tools at all;
• What verification obligations apply before an AI output informs a certificate or sign-off;
• Who carries the liability if an AI-assisted output is wrong; and/or
• Whether the professional's PI insurance actually responds to an AI-related claim.

That last point deserves particular attention. Professional indemnity policies in the construction sector weren't written with AI-assisted working in mind. Whether an existing policy responds to a claim arising from an AI-generated compliance report isn't a question with an obvious answer – and it's not a question you want to be asking for the first time after a defect has emerged.

It's not just the design team

This isn't just a concern for architects and engineers. Main contractors and specialist subcontractors are increasingly using AI tools for programming, sequencing, buildability analysis, and technical coordination. If your building contract and collateral warranties don't address AI tool use, you may have less recourse than you think if a contractor-side AI failure contributes to a defect or delay.

What to ask before your next appointment is signed

You don't need to prohibit AI – that would be neither practical nor desirable. But you do need your contracts to reflect how your professional team is actually working. Before signing your next set of appointments, it's worth asking:

• Does this appointment address AI tool use, and if so, how?
• What verification obligations does the professional have before an AI output informs a certificate?
• Does the professional's PI policy cover AI-assisted work, and have they confirmed that with their broker?
• If an AI-assisted sign-off is later challenged, what recourse do I have under this appointment?

Irish courts haven't yet ruled on these questions. When they do, the appointment document will be the starting point. The developers and investors who have already addressed this will be in a considerably stronger position than those who haven't.

If you would like us to review your professional appointments or advise on how to address AI governance in your construction contracts, I am happy to have that conversation.