Progressive Grounds

The RICS responsible-AI standard: what your firm must now do

By Ian Bester · Updated 15 October 2026 · Guidance, not legal advice

Short answer: RICS's professional standard Responsible use of artificial intelligence in surveying practice (1st edition, September 2025) has applied since 9 March 2026 (S1 §1.3). It binds RICS members and RICS-regulated firms in every jurisdiction (S1 §1.2). Where AI outputs have a material impact on a surveying service, firms must keep an AI register and a risk register, carry out supplier due diligence, record reliability decisions and tell clients in writing (S1 §3–§4). Nearly every operative duty is a "must", not a "should" (S1 p. 2).

Who it applies to

  • RICS members (individuals) and RICS-regulated firms, worldwide (S1 §1.2).
  • Contractors that employ RICS members. An MRICS commercial manager or in-house QS carries the member-level duties: knowledge (§2), reliability decisions (§4.2) and client disclosure (§4.3). The firm-level duties apply only if the contractor is itself RICS-regulated (S1).

What counts as "material" AI use

The standard covers only AI outputs that have a material impact on delivering the surveying service (S1 §1.2). RICS's own examples are:

  • a document summary relied on in a report
  • AI drafting all or a significant part of an opinion
  • AI recommending which part of a building to investigate

General tools count. In RICS's consultation, 53% of the 71 respondents said they already used generic tools such as ChatGPT or Copilot in surveying services (S2, pp. 2–3; a self-selected sample). AI embedded in software you already use is also in scope for due diligence (S1 §4.1).

The ten things a firm must have in place

  1. A record of which AI uses are material, and why (§1.2)
  2. A written assessment that AI is the right tool, covering seven factors including data risk, error and bias (§3.2)
  3. A written AI register: each material system, its purpose, first use and next review date (§3.2)
  4. A responsible-AI policy setting out roles, liabilities, at-least-annual training, human oversight and risk guidance (§3.2)
  5. An AI risk register with likelihood, impact, mitigation, appetite and a RAG rating, reviewed at least quarterly by the people who decide on AI (§3.3)
  6. Data controls, and no upload of confidential data to AI without the affected stakeholders' express written consent in advance, and only after checking the risk is acceptable (§3.1)
  7. Supplier due diligence in writing before procuring material AI, covering six required topics including the supplier's liability (§4.1)
  8. A written reliability decision for each material output, by or under a named, qualified surveyor, or regular dip samples for high-volume output (§4.2)
  9. Client disclosure in advance and in writing, with terms of engagement stating when AI is used, PI cover, contest, redress and opt-out (§4.3)
  10. A written information pack for clients who ask (§4.4)

How it is enforced

The standard sets out no separate penalties. RICS enforces it through its normal regulation and disciplinary process, and takes it into account when judging whether a member or firm acted with reasonable competence. A judge or adjudicator is also likely to take it into account (S1 p. 2).

How it fits with the EU AI Act, ISO/IEC 42001 and UK GDPR

  • EU AI Act: the standard names it as legislation that takes precedence (S1 §1.2). It can reach UK firms with EU work or EU suppliers. Take legal advice on your position.
  • ISO/IEC 42001: an AI management-system standard. The RICS standard doesn't mention or require it. Our view: ISO/IEC 42001 controls can evidence the RICS duties.
  • UK GDPR: the standard defines private and confidential information by reference to UK GDPR personal data. Where local law conflicts with the standard, the law takes precedence, and the conflict must be recorded and reported to RICS (S1 §1.2, glossary).

A ten-question readiness check

  1. Have we listed every AI tool, including embedded AI, and decided which uses are material?
  2. Is there a dated AI register?
  3. Is there a written "right tool" assessment?
  4. Is there an AI policy with roles, liabilities, training and oversight?
  5. Was the AI risk register reviewed by decision-makers this quarter?
  6. Do we get written consent before confidential data goes into AI?
  7. Is there written due diligence for each material AI supplier?
  8. Is there a reliability-decision template with a named surveyor, and a dip-sampling schedule?
  9. Do our terms of engagement cover AI use, PI, contest, redress and opt-out?
  10. Could we hand a client the §4.4 information pack tomorrow?

See where the board stands on technology control: take the construction scorecard. Read the official standard on rics.org (PDF); the consultation response records how the duties were set.

How Progressive Grounds helps

We start with a short, fixed-scope look at what AI is actually in use and which duties apply, then set up the register, risk register, policy and supplier questions in proportion to your firm. The board stays the owner.

AI Governance Decision Brief · AI Exposure Audit · Supplier AI Compliance Response · Fractional CIO for engineering and construction firms

Questions

When did the RICS AI standard come into force?

It has applied since 9 March 2026. The first edition is dated September 2025, and RICS published it online on 17 November 2025 (S1 §1.3; S3).

Does using ChatGPT or Copilot bring us into scope?

It can. The test is whether the output has a material impact on the surveying service, for example a summary relied on in a report (S1 §1.2).

Is there a penalty for non-compliance?

There are no specific sanctions. RICS applies its normal regulatory and disciplinary process, and courts are likely to consider the standard when judging competence (S1 p. 2).

Do we need ISO/IEC 42001 certification?

No. The RICS standard doesn't require or mention it.

Book a 20-minute call Take the construction scorecard

Sources

  1. S1: RICS, Responsible use of artificial intelligence in surveying practice, professional standard, 1st edition, September 2025, effective 9 March 2026. PDF.
  2. S2: RICS, Response to consultation on the Responsible use of AI professional standard, 1st edition, August 2025 (71 respondents).
  3. S3: RICS, Responsible use of AI landing page (published 17 November 2025).