FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Building control approvers
AI rule-checkers can flag possible non-compliance in drawings, but the plans assessment decision stays with a registered, competent building inspector. A clean AI report that misses a fire safety or structural breach is still the inspector's responsibility.
Part of: AI in building control approvers
In short
Software that reads drawings and specifications against the Approved Documents can speed up plans assessment, but it cannot make the decision. Plans assessment is a restricted activity, and RBI code 3.1 limits it to inspectors registered and competent for the work, while 4.2 makes them accountable for decisions and Annex 1.2 requires auditable reasoning. The main risk is false assurance: a tool that does not flag a Part B or structural problem can lead to a lighter review. PI usually treats a missed breach as negligence whether or not AI was used, subject to the wording.
Plans assessment means checking that proposed work, as shown in drawings, specifications and calculations, will comply with the Building Regulations. AI tools now offered to building control bodies can extract room sizes, travel distances, door widths and fire compartment lines from drawings, compare them with the Approved Documents and list apparent gaps. Generative AI is also used to summarise specifications and draft the schedule of conditions or further information requests.
These tools can be useful for consistency, especially on repetitive domestic work. They are weaker where compliance depends on judgement: alternative fire engineering approaches, unusual structures, interactions between parts of the regulations, or information that is missing rather than wrong.
Several existing duties apply to this work:
The RBI code of conduct and operational standards rules do not mention AI, but nothing in them allows the assessment decision to be delegated to software. The BSR codes and AI guide sets out each provision.
The common failure is not a tool that invents a breach, but one that misses one. Rule-checkers work from what they can read. They may not recognise a fire door shown in a schedule rather than on plan, a cavity barrier detail referenced in a separate document, or a structural element whose adequacy depends on a calculation they have not been given. A green result can then shorten the human review that should have caught it.
Submissions themselves may be AI-generated. CROSS-UK Report 1554 described an AI-written calculation reaching the design stage without competent verification, and the panel suggested approving authorities may wish to ask how far AI contributed to submissions. Two layers of AI with no competent human check is the scenario to avoid.
Illustrative: an RBCA uses a rule-checker on a mixed-use scheme. The tool reports compliant escape widths and flags nothing on compartmentation. The inspector, under time pressure, approves with standard conditions. During construction it emerges that a protected stair was shown with an inadequate fire separation from a commercial unit. Redesign causes delay, and the developer alleges the approval was negligent. The file contains the AI report but no note of the inspector's own check of the stair.
A claim that plans were negligently approved is usually treated as professional negligence, subject to the policy terms; the use of software alone should not change that. Cover issues more often arise from fire safety and cladding restrictions, which may limit or exclude claims about fire performance, and from late notification. For higher-risk buildings, where BSR is the building control authority, the stakes and the documents are larger; see the higher-risk building regime and AI review of Gateway 2 applications. The BSR codes require RBCAs and self-employed RBIs to be suitably insured but set no minimum limit.
Useful controls include:
Apex arranges PI for building control approvers and inspectors. A broker can review your cover and how your policy treats AI-assisted plan checks, including fire safety restrictions. More guides are on the AI in building control hub.
No. Plans assessment is a restricted activity for RBIs registered and competent for the work (RBI code 3.1), and the inspector is accountable for the decision (4.2). An AI tool can support the review, but the decision and its recorded reasoning must be the inspector's.
Keep the tool output, the version and scope of the tool, and your own note of what you checked and why you reached your decision. Annex 1.2 of the RBI code requires records that show how decisions were reached, and OSR 1.7 requires records to be kept for at least 15 years.
Not usually for that reason alone. A missed breach is commonly treated as negligence, subject to the policy terms. Fire safety restrictions, AI clauses in some wordings and whether you disclosed your AI use at renewal are more likely to affect the outcome.
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.
