> ## Documentation Index
> Fetch the complete documentation index at: https://help.guardbook.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# The RICS AI standard: what it is and what it requires

> An introduction to Responsible use of artificial intelligence in surveying practice, its five sections, and the obligations each one creates.

*Responsible use of artificial intelligence in surveying practice* takes effect on **09/03/2026**. It is a professional standard, so compliance is not optional for RICS members and RICS-regulated firms.

## What kind of standard it is

It is a **conduct** standard. It does not tell you how to use a particular kind of AI for a particular kind of surveying work, and it does not approve or prohibit specific tools. It sets out how a professional should behave when AI is involved in the work.

It applies globally. Where it conflicts with local legislation the legislation wins, but the firm must record the conflict and report it to RICS.

## The materiality test runs through everything

Most of the substantive duties are switched on by **material impact**: whether an AI output is capable of influencing the delivery of the surveying service and, if so, how, on the facts of the particular use.

The standard's own examples of material outputs are a summary of a document that is then relied on in writing a report, text composing all or a significant part of an opinion, and a recommendation about which part of a building to investigate.

> [!NOTE]
>
> If a firm determines its AI use **will** have a material impact, it must record that determination and the reasoning behind it. The standard does not require a record of a negative determination.
>
> So "you must record whether AI has a material impact" overstates the duty. The record is required when the answer is yes.

## The five sections

> [card] Section 1. Introduction
>
> Scope, the materiality test, and how the standard interacts with local law.

> [card] Section 2. Baseline knowledge
>
> Individual competence. Members using AI must understand the types of AI system and how they work, the risk of erroneous output, the inherent risk of bias, and the data risks. This is personal knowledge, not firm-run training.

> [card] Section 3. Practice management
>
> Firm-level obligations. **3.1** data governance, **3.2** policies and system governance, **3.3** the risk register.

> [card] Section 4. Using AI
>
> The operational core. **4.1** procurement and supplier due diligence, **4.2** reliability decisions on outputs, **4.3** terms of engagement and client communication, **4.4** explainability on request.

> [card] Section 5. Development
>
> Applies only to firms that develop AI in house. They must apply the standard, as relevant, to the development process too.

See [the clauses](https://help.guardbook.ai/rics-standard/clauses.md) for what each one requires in practice, and [who it applies to](https://help.guardbook.ai/rics-standard/who-it-applies-to.md) for scope.

## Three duties firms most often miss

* **Training is three duties, not one.** Section 2 individual knowledge, clause 3.1 annual training for staff with access to private and confidential data, and clause 3.2 annual training for everyone involved in AI use. Doing one does not discharge the others.
* **The risk register has to be operated.** Clause 3.3 requires review and update at least quarterly, not creation once.
* **A negative reliability conclusion has to reach the client.** Clause 4.2 requires it in writing, with your reasoning or a summary. Recording it internally is not enough.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.