Aipedia: Architects

AI in architecture: rules, risks and professional indemnity

Aipedia, by the Apex broking team · Last reviewed 2026-10-27

Most UK architectural practices now use AI, but no architecture regulator has issued AI rules, so the existing ARB Code, RIBA codes and building safety law decide what is acceptable. That matters because the architect, not the software, carries the professional duty and the PI claim.

Part of: AI in Architects

In short

RIBA's 2026 AI Report found 74% of practices use AI on at least some projects, yet only 19% have a documented AI policy. ARB has published no AI guidance, so the Architects Code in force from 1 September 2025 applies as written: honest promotion, work within competence, supervision, written terms, accurate records, confidentiality and adequate PI. CROSS-UK Report 1554 shows what happens when an AI-written calculation goes unchecked. Architects' PI usually responds to negligent professional work whatever tool was used, subject to the policy terms and full disclosure of how AI is used.

How UK architects use AI

Adoption has moved quickly. RIBA's AI Report 2025 found the share of practices using AI rose from 41% in 2024 to 59% in 2025. The RIBA AI Report 2026, based on a survey in March and April 2026, puts it at 74% of practices using AI on at least some projects, with 31% using it on every or most projects (6% in 2024).

Among practices that use AI, the main tasks reported were:

early-stage design visualisation (about 56%);
compliance checking (40%);
practice management, such as drafting reports and documentation (32%);
specification writing or technical design (24%);
product and material selection (18%).

The same report shows the profession is uneasy. 76% agree AI increases the risk of their work being imitated, and 55% agree it will let people without sufficient qualification design buildings. Only 19% of practices have a documented AI policy, with 17% developing one and 15% planning one.

What ARB and RIBA require

The Architects Registration Board, the statutory regulator under the Architects Act 1997, has no AI-specific guidance as at October 2026. The Architects Code of Conduct and Practice, in force from 1 September 2025, does not mention AI, technology or digital tools. It still governs every AI-assisted task. The standards that bite hardest are:

1.2 promote services responsibly and accurately, which covers AI-generated imagery and claims;
3.1 work within the limits of competence, and 3.2 keep up to date with current standards;
3.5 appropriate supervision and sufficient resources;
4.1 clear written terms before work begins, 4.5 accurate records and 4.6 confidentiality and security of information;
4.7 adequate and appropriate professional indemnity insurance.

RIBA's Code of Professional Conduct and Code of Practice for Chartered Practices (both 1 April 2021) also predate generative AI, but require competent work, no misleading statements "through words or images", confidentiality, data protection compliance, written terms, records and suitable insurance. RIBA says it will publish an AI Overlay to the RIBA Plan of Work and a policy for responsible use of AI during 2026. The ARB and RIBA guide maps each standard onto AI use.

Building safety and the CROSS 1554 warning

In England, Part 2A of the Building Regulations 2010 puts competence duties on designers (regs 11F to 11H) and requires a designer to ensure the design would comply if built (reg 11K). An AI tool cannot hold that duty.

CROSS-UK Report 1554 (16 July 2026) is the clearest UK warning so far. An architect submitted a calculation report, which the reporting structural engineer was confident was largely AI-written, to justify heavy rooftop plant on an existing flat-slab concrete building with retail below and flats above. It contained fundamental errors and an overloading risk, no competent structural engineer had verified it, and the architect dismissed the engineer's comments. The CROSS panel pointed to the Part 2A competence duties and said designers should disclose AI-assisted outputs in engineering submissions.

No ARB disciplinary decision or UK court judgment about an architect's use of AI had been verified as at October 2026.

The main risks

Misleading images: generative renders that show more than the design, budget or planning position can deliver.
False assurance: compliance checkers that miss a fire safety, access or overheating issue.
Outdated or wrong content: AI-drafted specifications, statements and calculations that cite superseded standards or invented policy.
Work outside competence: using AI to produce engineering or specialist outputs the practice cannot check.
Records and confidentiality: AI notetakers that mis-record instructions, and client information pasted into public tools.
Copyright and imitation: uncertainty over who owns AI-generated design options.

How architects' PI responds

Standard 4.7 makes PI compulsory for registered architects. ARB's PII guidance (19 June 2025) sets a minimum of £250,000 each and every claim, allows aggregate cover for fire safety and cladding, asbestos and pollution if sized for multiple claims, and requires run-off of at least six years (five in Scotland). Restricted cover is acceptable only after reasonable efforts to find full cover. The guidance does not mention AI.

Most architects' policies are written on a civil liability basis and usually respond to negligent professional work regardless of the tool involved, subject to the policy terms. Silence on AI is not the same as a guarantee, though. Fire safety restrictions, exclusions, the duty of fair presentation and the question of whether work was within the declared profession all still apply. RIBA lists PI and liability for AI use as unresolved questions for the profession and insurers. See the architects' PI guide for 2026 and whether PI covers AI mistakes.

Practical controls

Write a short practice AI policy covering approved tools, data rules and review.
Have a competent person check every AI output that leaves the practice or informs design.
Never let AI produce work outside the practice's competence, such as structural calculations.
Label AI-generated images and declare AI use where planning or approval rules require it.
Keep records of what was generated, what was checked and by whom.
Tell your broker how the practice uses AI at renewal.

Aipedia guides for architects

ARB and RIBA on AI: how the Code and RIBA codes apply to AI use.
AI renders and visualisations: misleading images and planning declarations.
AI building regulations checks: false assurance on Parts B, M and O.
AI specification writing: outdated standards and copied text.
AI and structural calculations: lessons from CROSS 1554.
AI design and access statements: the Planning Inspectorate declaration.
Generative design and copyright: ownership and imitation.
AI feasibility and massing tools: yields relied on for land deals.
AI daylight, energy and overheating studies: performance-gap claims.
AI notetakers for meetings: records, consent and disputes.
AI-drafted fee proposals and appointments: scope gaps and liability terms.
AI chatbots and marketing: responsible promotion.
AI product and material selection: substitutions and fire performance.
Does architects' PI cover AI mistakes?: wordings and disclosure.
Practice AI policies and telling clients: what to write down.

How Apex helps

Apex arranges PI for architects and practices of all sizes. A broker can review your cover against the ARB minimum terms and explain how your policy is likely to treat AI-assisted design, documentation and visualisation work.

Sources

Frequently asked

Does ARB allow architects to use AI?

Yes. ARB has published no AI guidance and does not prohibit AI. The Architects Code in force from 1 September 2025 applies to AI-assisted work in the same way as any other work, so the architect remains responsible for accuracy, competence, supervision, confidentiality and records.

How many architects use AI?

RIBA's AI Report 2026 found 74% of practices use AI on at least some projects and 31% on every or most projects. Visualisation is the most common use, followed by compliance checking, practice management, specification writing and product selection.

What was CROSS-UK Report 1554?

A July 2026 safety report about an architect who submitted an apparently AI-written calculation to justify heavy rooftop plant on an existing building. It contained fundamental errors and had not been checked by a competent structural engineer. The panel cited the Part 2A competence duties.

Will my PI insurer ask about AI?

A proposal form may or may not ask about AI. Either way, the duty of fair presentation means material changes in how the practice works, including heavy reliance on AI for design or technical outputs, should be disclosed to insurers at renewal.

Check your PI covers how you use AI

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 callback

Apex 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.