FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Apex Insurance Brokers
Speak to a brokerGet a quote →

Aipedia: Architects

Practice AI policies and telling clients about AI

Every practice using AI should have a short written AI policy and a clause in its appointments explaining how AI is used. Only 19% of practices had a documented policy in 2026, yet it is the easiest evidence of control for clients, ARB and insurers.

In short

RIBA's AI Report 2026 found 74% of practices use AI but only 19% have a documented AI policy, with 17% developing one and 15% planning one. No ARB or RIBA rule yet requires a policy, though RIBA plans a responsible-use policy in 2026. A practice policy turns Code Standards 3.5, 4.5 and 4.6 into everyday rules. An appointment clause meets Standard 4.1 and avoids disputes with clients who restrict AI. Both help show insurers how AI risk is managed.

Why a written policy matters

The RIBA AI Report 2026 found most practices use AI but few have written down how. RIBA's response to the report flags professional indemnity, liability, intellectual property, data security and accountability as unresolved issues, and RIBA plans an AI Overlay to the Plan of Work and a policy for responsible use of AI in 2026. ARB has published no AI guidance.

No rule yet requires a practice AI policy. But the Architects Code expects supervision (3.5), accurate records (4.5), confidentiality (4.6) and clear written terms (4.1), and a policy is the simplest way to show those duties are met when AI is involved. See the ARB and RIBA guide.

What a practice AI policy should cover

  1. Approved tools and who can approve new ones.
  2. Permitted and prohibited uses, for example no structural calculations or fire safety conclusions from AI.
  3. Review: who checks AI outputs, by task and risk level.
  4. Data rules: what client, personal or security-sensitive information must never go into which tools.
  5. Images: labelling AI visuals and planning declarations.
  6. Records: what to keep about prompts, outputs and checks.
  7. Client terms: how AI use is explained and how client restrictions are followed.
  8. Training and review: who is trained and when the policy is updated.

Keep it short enough that staff read it. Two or three pages is often enough for a small practice. Name a person responsible for the policy, and make sure new starters and consultants working on the practice's behalf see it, since Standard 3.4 requires architects to check the competence of anyone engaged on their behalf.

Telling clients and appointment clauses

Standard 4.1 requires clear written terms before work begins, and Standard 5.3 requires managing expectations. An appointment AI clause might cover:

Some clients and contracts restrict or ban AI use, or require disclosure. Read schedules and framework terms for these, as an AI-drafted summary may miss them. See AI-drafted fee proposals and appointments.

Data protection points

Client briefs, drawings and personal data put into AI tools engage Code 4.6 and RIBA Code 4.1 to 4.7, as well as UK GDPR. If the practice uses AI to make significant decisions about individuals, such as staff or applicants, the automated decision-making rules inserted by the Data (Use and Access) Act 2025 s.80, in force 5 February 2026, also apply. Meeting recordings are covered in AI notetakers.

Illustrative scenario

Illustrative. A practice wins a public sector framework place. The framework terms prohibit uploading project information to external AI tools without consent. A team member uses a public AI tool to summarise a confidential brief. The client discovers this and raises a breach of contract and confidentiality complaint. The practice has no AI policy and cannot show what was uploaded or why.

Documenting AI use for insurers

Insurers may want to know how AI is controlled. A written policy, a list of approved tools and records of review help at renewal and when defending a claim. Keep evidence of training and of policy updates. See what to document about AI for PI and does architects' PI cover AI mistakes.

How Apex helps

Apex arranges PI for architects. A broker can explain how insurers view practice AI policies and help present your controls at renewal. See also the AI in architecture overview.

Sources

Frequently asked

Does ARB require a practice AI policy?

No. As at October 2026 ARB has no AI guidance. A written policy is still the clearest way to show you meet Code standards on supervision, records and confidentiality when using AI, and RIBA plans its own responsible-use policy in 2026.

Must I tell clients I use AI?

No rule specifically requires it, but Standards 4.1 and 5.3 point towards explaining it in your terms. Some client contracts require disclosure or restrict AI, and those terms must be followed.

How long should a practice AI policy be?

Short enough to be read and followed. For many small practices, two or three pages covering approved tools, banned uses, review, data rules, images, records and client terms is enough. Review it regularly as tools and guidance change.

Will an AI policy reduce my PI premium?

There is no guarantee. A clear policy and good records help insurers understand the risk and support a fair presentation, and they make claims easier to defend. Pricing depends on many factors.

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.