Is AI safe for survey reports?

AI can be used safely for survey reports when two things are true: the AI supports the surveyor's professional judgement rather than replacing it, and client data is handled carefully. The biggest professional worry is that an AI, left to its own devices, writes something that leaves the surveyor exposed: stating a defect as definite when the reality is nuanced, or describing something that was never observed. Managed properly that risk is controllable, and the surveyor remains the author who signs the report off. Since 9 March 2026, RICS has also made responsible AI use a formal requirement through its professional standard.

This page explains the professional indemnity concern and how to manage it, what the RICS standard requires, what good data handling looks like, and how Blooming Surveys is built to meet all of it.

The professional indemnity concern, and how to manage it

The specific fear for surveyors is that AI will write something that exposes them to a claim, and it is a fair concern. Left unconstrained, AI has two tendencies that are dangerous in a survey report:

• False confidence. AI tends to sound definitive and authoritative even when the truth is uncertain or nuanced, which is the wrong tone for a report that has to stand up under professional indemnity.
• Hallucination. AI can state things that were never observed, inventing detail that is not grounded in the inspection.

A tool that is safe to use under PI has to manage both. That means writing at the surveyor's own level of confidence, with the caveats they would normally apply, rather than imposing a false certainty, and grounding every statement in the facts the surveyor actually recorded, rather than letting the model fill gaps with invention. And the surveyor must remain the author who reviews and signs off, so nothing reaches the client unchecked.

What the RICS AI standard requires

The RICS professional standard "Responsible use of artificial intelligence in surveying practice" is mandatory for RICS members and regulated firms from 9 March 2026. It applies whenever AI has a material impact on a surveying service, which includes using AI to draft significant parts of a report. The duties sit with the surveyor and the firm, not with the software, but the tools they choose make those duties easier or harder to meet. In summary, a surveyor using AI must:

• Keep professional judgement and stay accountable. A named surveyor must apply professional judgement and scepticism to AI output and remain responsible for it. For high-volume or automated use, the firm should run regular dip-sample checks rather than rely on the AI unchecked.
• Govern confidential data. Firms must store private and confidential data securely, for example using encryption and/or backups, restrict who can access it, train staff, and prepare data in a way that protects privacy. Critically, they must not upload private and confidential data to AI systems without express written consent from the people affected.
• Do due diligence on AI suppliers. Before using a third-party AI system, the firm must obtain and record written information from the supplier, covering compliance with data and confidentiality law, the training data and any bias, environmental impact, and the supplier's liability.
• Tell clients, in writing. Terms of engagement must set out when and where AI is used, the extent of any PI cover for AI, how a client can contest its use or seek redress, and how they can opt out.
• Be able to explain it. On request, the firm must be able to explain the type of AI used, how it works and its limits, the due diligence done, and the checks on the reliability of its output.
• Manage the risk. Firms must keep a risk register covering AI risks such as bias, erroneous output, and data retention, and review it regularly.

"Private and confidential data" here means information that must not be unlawfully disclosed, including personal data (anything identifying a living person) and commercially sensitive data. For a survey, that is mainly the client's identity and the property address, plus any personal details that find their way into notes.

What good data handling looks like for any AI survey tool

Whatever tool a surveyor uses, the same questions decide whether it is safe under PI and the RICS standard. A well-built tool should:

• Keep client identifiers, such as the client's name and the property address, out of what is sent to the AI.
• Be clear about what is sent to AI providers and what those providers do with it, including whether the data is deleted after processing.
• Store private and confidential data securely.
• Write at the surveyor's own level of caution and stay grounded in their recorded facts, rather than adding false confidence or invented detail.
• Leave professional judgement, the condition ratings, and the sign-off with the surveyor.
• Be transparent about where data is stored, including any processing outside the UK or EU.

Anyone evaluating a tool should read its privacy policy and confirm these points before uploading real reports.

Is a general AI chatbot like ChatGPT safe for survey reports?

Pasting survey notes into a general chatbot such as ChatGPT is a different proposition. There is no separation of client identifiers, your input may be used in ways you do not control, and there are no survey-specific safeguards, which makes it hard to meet either your PI obligations or the RICS standard. Purpose-built survey software is designed around these risks. For how the approaches differ, see how AI survey report software works.

How Blooming Surveys does this

Blooming Surveys is built around both halves of the question, your judgement and your data.

Written to your level of confidence

The PI risk of false confidence is handled at the source: Blooming Surveys learns how you write from your previous reports, including the caveats and the degree of certainty you normally use, so the draft mirrors your professional tone rather than an AI's false authority. It can apply cautious, PI-aware language, for example "likely caused by" rather than a definitive statement.

Grounded in your facts

The hallucination risk is handled by grounding: the draft is built only from the inspection notes you recorded, which keeps it tied to the facts and guards against the AI inventing detail. It will not invent defects.

You stay the author

You note the condition rating on site and the software uses it, or suggests one for you to confirm or change if you leave it blank. Every line is editable, nothing is locked, and you sign off personally, so the report is your professional opinion and nothing reaches the client unchecked.

What is and is not sent to the AI

You enter the client's name and the property address in dedicated fields, and that information is not sent to the AI. Your notes and photos are sent to a third-party AI provider to draft the report. The provider does not use the data we send to train its models, and retains it for up to 30 days for abuse monitoring before deleting it, unless it is legally required to keep it longer. Because notes are sent as written, we do not strip personal details out of free text, so we ask you to keep client and third-party personal details out of your notes and to use the dedicated fields for names and addresses.

Where your data is stored

Your data is stored primarily on EU-based servers, and may occasionally be processed or stored outside the EU, including in the US, with Standard Contractual Clauses in place. Blooming Surveys retains your reports and notes in your account so you can access them, and you can request access, correction, or deletion at any time. Blooming Surveys Ltd is registered with the ICO. Your content and writing style are yours and are not shared with other surveyors.

Built to support the RICS AI standard

Because you remain the named author who confirms findings, ratings, and sign-off, Blooming Surveys supports the standard's core requirement that professional judgement stays with you. Keeping the client name and address out of the AI, and asking you to keep personal details out of notes, helps you avoid uploading confidential data to an AI system. And we provide the supplier information you need for your due diligence and explainability obligations. The standard still places duties on you that no tool can discharge, such as telling clients in writing that AI is used, offering opt-out and redress in your terms of engagement, and keeping your own risk register. To help with that last one, we have a ready-made AI risk register template: ask us for a copy.

A note for the cautious

If you are weighing AI for survey work under PI, the right approach is to read the provider's privacy policy, check what is and is not sent to the AI, confirm retention and deletion terms, make sure the tool helps you meet the RICS AI standard, and keep professional judgement and sign-off with the surveyor. We have set out our position above; our full privacy policy and terms of service have the detail.

See also how it works, and the Level 2 and Level 3 pages.

Frequently asked questions