The standard and your firm: knowledge, registers and suppliers

The written material

A mandatory conduct standard

Responsible use of artificial intelligence in surveying practice is an RICS professional standard, first edition, published in September 2025 and effective from 9 March 2026. It sets requirements for members and regulated firms in all jurisdictions. Where it conflicts with legislation, the legislation takes precedence, and the conflict must be recorded and reported to RICS.

It is supportive of AI. It is also clear that the skill and experience of the surveyor sit at the centre, and that the profession must guard against complacency. In the words RICS used when it came into effect: AI assists professional practice, it does not replace it, and the surveyor remains accountable for every piece of professional advice.

Material impact: when the standard applies

The standard applies to the outputs of AI systems that have a material impact on the delivery of the surveying service. That depends on whether the output is capable of influencing the service and, if so, how.

Its examples are practical. An output summarising documents that are then relied on in a report. An output composing all or a significant part of an opinion. An output recommending which part of a building to investigate for a fault. A tool that tidies your diary is not in the same position.

If you determine that your use of AI will have a material impact, you must make a record of that determination and the reasoning behind it.

Knowledge and data

Members who use AI to deliver surveying services must develop and maintain sufficient knowledge. As a minimum that means a basic understanding of the types of AI system and their limitations and failure modes, the risk of erroneous output, the inherent risk of bias, and data usage and data risks.

Firms must safeguard private and confidential data: stored securely, access restricted to staff who strictly need it, staff trained at least annually, and data prepared in a way that protects privacy. They must refrain from uploading private and confidential data to AI systems, except where there is express written consent in advance from affected stakeholders and the firm has taken reasonable steps to satisfy itself that the upload does not pose an unacceptable risk.

The register, the policy and the risk register

Before using an AI system with a material impact, a firm must carry out and record in writing an assessment of whether AI is the most appropriate tool, considering the alternatives, environmental and stakeholder impact, data risks, and the risk of error and bias. A standing written policy can serve as that assessment.

The firm must keep a written register of each such system: what it is, the purpose it is used for, the date it was first used, and the date its use will next be reviewed. It must have responsible use policies setting out roles and liabilities, training at least annually, and how human judgement interacts with the tool.

And it must run a risk register covering bias, erroneous outputs, limits to what is known about the system and its training data, and the retention or use of data the firm puts in. Each risk carries a description, likelihood and impact, a mitigation plan, the firm's risk appetite, status updates and a red, amber, green rating or similar. It must be reviewed and updated at least quarterly.

• AI register: system, purpose, date first used, next review date
• Responsible use policy: roles, annual training, human control
• Risk register: reviewed at least quarterly

Due diligence before you buy

Before procuring an AI system with a material impact, a firm must carry out detailed due diligence: request information from the supplier in writing, follow up in writing, record and assess what comes back, and keep a record of how far the system was tested for fitness for purpose.

The written requests must cover, as a minimum, the environmental impact of the system, the stakeholders involved in developing it, compliance with data and confidentiality laws, permissions for any data about individuals, the accuracy, relevance and diversity of the training data including known gaps and bias, and the type and extent of the supplier's liability.