Can AI Write a Flood Risk Assessment?

Ask any chatbot to write a flood risk assessment and it will hand you a confident, well-formatted document in seconds. That is the easy part. It is also the misleading part, because a flood risk assessment is not really a writing task. It is an evidence task, and the words are only the wrapper around survey data, modelling, current policy and professional judgement.

So, can AI write a flood risk assessment? It can produce text that resembles one, but not a report a planning authority will accept. An FRA is judged on whether its data is current, whether it covers every source of flooding, whether the site has actually been assessed, and whether a competent person will put their name to it, and on all four counts an AI-generated document fails by design rather than by accident. The polish is real. The evidence is not.

Around 6.3 million properties in England now sit in areas at risk of flooding, according to the Environment Agency, so scrutiny of flood risk assessments is tightening rather than easing. A report that cannot survive that scrutiny is a false economy.

Why writing is only the last 10% of a flood risk assessment

Most of the work in a flood risk assessment for planning happens before a single sentence is written. A language model can produce the write-up in seconds. It cannot produce the evidence underneath it, and an assessment with a polished write-up over thin or wrong evidence is exactly the kind a statutory consultee sends back.

The order of the job runs roughly like this:

1. Gather current flood data for the specific site, from every source, at the resolution planning requires.
2. Screen each source of flooding, from rivers and sea to surface water, groundwater, sewers and reservoirs, and rule each in or out on evidence.
3. Build or commission hydraulic modelling where the published maps are not detailed enough for the site.
4. Design mitigation that actually works: finished floor levels, safe access and escape, and a workable drainage strategy.
5. Apply national and local policy to the specific scheme, and only then write it up.

The writing is the final step and the smallest part. Everything that makes the report defensible sits in the four steps before it.

What a planning authority actually checks in an FRA

The Environment Agency, as statutory consultee, and the lead local flood authority do not grade an assessment on how well it reads. They test it against specific requirements, and these are the points where an AI-written report tends to come apart.

The current framework is explicit that the starting point is every source of flooding, rather than the obvious one:

Against that, a reviewer checks a short but demanding list:

• Current data and allowances: the maps and figures must be the live ones. The Flood Map for Planning was updated on 28 May 2026 with surface water climate change extents on the upper-end 2070s allowance, and climate change allowances vary by catchment and epoch.
• Every source of flooding: since the September 2025 PPG update, carried into the August 2026 NPPF, the assessment must consider surface water, groundwater and sewers as well as rivers and the sea.
• A genuine sequential test: and, where needed, an exception test, argued for the actual site. The Yatton judgment shows how finely the planning balance can turn.
• Lifetime safety: the development must stay safe for its lifetime with no increase in flood risk elsewhere, evidenced by finished floor levels, safe access and a drainage strategy that works.
• A competent, accountable author: someone qualified and insured who will stand behind the conclusions, including at appeal.

Where an AI-generated flood risk assessment fails those checks

Mapped against that list, the gaps are not stylistic. They are structural. A model reaches for figures from its training material, defaults to the flooding that dominates the text it has read, and cannot survey a site or carry professional liability. The table below sets each check against how an AI-only report typically fails it.

Will the Environment Agency accept an AI-written FRA?

You can submit one, but it is unlikely to be accepted. The Environment Agency and the lead local flood authority test the substance, not the wording, and because an AI-generated report typically uses out-of-date data, misses surface water or groundwater, and has no competent author to stand behind it, the realistic outcome is an objection or a holding direction rather than a validation. It clears the reviewer's desk slowly, if at all.

Nothing in the process detects "AI" as such, and nothing needs to. A consultee does not have to prove how a report was written. It only has to find that the evidence is wrong, thin or missing, and an AI-only report gives it plenty to find. Whether a person, a template or a model produced the text is beside the point. The report stands or falls on whether it is right.

What happens if you submit an AI-only FRA?

The practical cost lands later, not at submission. A report that misses a source of flooding or uses superseded allowances draws an objection or a holding direction from the consultee. That means delay while it is reworked, and in the worst case a refusal that has to be fought at appeal with the assessment redone properly anyway.

• Delay: an objection stalls the application while the report is corrected and re-consulted, often adding weeks or months.
• Cost: the saving from skipping a proper FRA is wiped out once fresh surveys, modelling and a re-submission are needed.
• Refusal and appeal: a weak flood case can sink an otherwise sound scheme, and appeals since April 2026 place more weight on first-time completeness.

Where AI genuinely helps in flood risk work

None of this makes AI useless to the people preparing assessments. It changes what it is for. In a consultant's hands it is a reasonable drafting aid: structuring a first draft, condensing long policy into plain English, or explaining a term to a client. The rule that keeps it safe is simple. Treat anything it produces about flood levels, allowances, modelling or policy as a claim to be checked against the primary source, never as a finding.

The judgement stays with the engineer. AI speeds up the typing around it, and nothing more.

At the system scale, AI is already doing real work, separate from the question of writing reports. Machine learning is combined with meteorological modelling to improve flood forecasting lead times, and Northumberland County Council's FloodAI project, funded through DEFRA's £200 million Flood and Coastal Innovation Programme, uses it for earlier flash-flood warnings. We look at this in full in how AI could improve flood risk management. In every credible case the pattern is the same: AI supports the system, and a person remains responsible for the decision.

Frequently asked questions

Talk to a flood risk consultant who can stand behind your FRA

If your application needs an FRA, the question that matters is not who, or what, wrote it. It is whether the report stands up to the people who review it. Talk to Unda's flood risk and drainage consultants for an assessment built on current data, real site evidence and policy applied to your scheme, signed by someone accountable for it.