An 'AI-Generated' Chart in the Hillhouse Contamination Report
One of the two panels in the Phase 3 report's key concentration-by-direction chart carries the caption 'Generated using AI'. The report never says what that means or how the panel was produced. That is a fair question to put to WSP and the Environment Agency, not a claim that anything is wrong.
From the Environment Agency’s Phase 3 soil investigation of the Hillhouse site.
The chart described below can be viewed directly in the public version of the WSP Phase 3 Factual Report, the soil investigation into PFOA and PFAS contamination around the AGC Chemicals Europe (formerly ICI) plant at Hillhouse, Thornton-Cleveleys, published by Wyre Council: wyre.gov.uk/downloads/file/2452/phase-3-factual-report, page 33.
One of the two central charts in the Phase 3 soil report carries a caption that never gets explained: “Generated using AI.” Nothing before it or after it says what that means.
Two panels, one measurement, two captions
Insert 4-2, on page 33 of the public report, presents two radar scatter plots side by side, titled “Radar Scatter Plots - PFOA Concentration in S(µg/kg) with Orientation (˚) of AGCCE facility – Combined Phase 2 and 3 Residential Soils Datasets (including (LHS) and Excluding (RHS) Poultry Keeper Property data)” (p.33). Both panels plot the same measurement, PFOA concentration against compass bearing from the site, on the same axes. The left panel includes every residential sample; the right panel excludes poultry-keeper properties.
Underneath the left-hand panel, the caption reads “PFOA concentrations in µg/kg (dry weight)”. Underneath the right-hand panel, in the same position, it reads “Generated using AI” (p.33).
What the report never tells you
Nothing in the surrounding text explains what “Generated using AI” refers to. It could describe the software used to render that one panel. It could describe a step in which the chart, or the data behind it, was produced or regenerated using an AI tool rather than plotted straight from the sample results. Those are very different things, and the report does not say which one it means, or why the caption sits under one panel and not the other showing the same measurement.
Why this is a question, not an accusation
This is not a claim that the right-hand panel is wrong, or that any underlying figure has been altered. Visually, the two panels look near-identical, consistent with the only stated difference being the poultry-keeper subset. But Insert 4-2 is one of the report’s central pieces of evidence. It is the chart that identifies the two highest PFOA concentration readings the combined dataset finds anywhere, at bearings of south-east and west-north-west from the site. An unexplained caption on a chart that important, in a report used to inform regulatory decisions about contamination near people’s homes, is worth asking about rather than assuming either way.
The question we are putting to WSP and the Environment Agency
Was the right-hand panel of Insert 4-2 plotted directly from the same combined dataset as the left-hand panel, using the same method? If an AI tool was used to generate or regenerate the chart, what was checked against the source data before it went into the report?
We are asking these openly, and putting them directly to WSP and the Environment Agency. Any response will be published here.
How we know what we know
Everything above comes from one public document: the WSP Phase 3 Residential Soils Investigation, commissioned by the Environment Agency and published by Wyre Council in March 2026 (link above, page 33). The chart and its two captions are Insert 4-2; the text introducing it is on page 32. We have not altered, re-plotted or reprocessed anything. We are pointing at a caption that is already there, on a public page, and asking what it means.