Fact check: Fake ‘train attack’ imagery and child obesity stats

This roundup of claims has been compiled by Full Fact, the UK’s largest fact checking charity working to find, expose and counter the harms of bad information.

Grok and Google Lens AI overviews claim fake imagery shows train attack

Grok and Google Lens AI overviews have been claiming that fake imagery shows the recent Huntingdon train attack.

Grok, the AI chatbot created by Elon Musk’s start-up xAI, told users one image of an injured man lying in a carriage was not AI-generated.

But garbled text on the backs of the police officers and paramedics and a stylised filter are clear signs that the picture was created with AI. The image also depicts a train with a different style of seating to the real train involved in the attack, while an account which posted it on X appeared to confirm it was created with AI.

When Grok was asked if the image was AI-generated, it replied: “No, this image appears to be a genuine photo capturing the aftermath of the stabbing incident on a Doncaster-to-London train stopped in Huntingdon, Cambridgeshire.”

Grok also appeared to double down on its verdict when questioned why ChatGPT said it was AI-generated, though it later acknowledged to one user the image was indeed AI.

An AI overview from the picture-searching tool Google Lens claimed without evidence that the same image was a “still from a BBC News report”, and linked to a BBC News article about the Huntingdon attack which did not feature the image. On another occasion it told us “this image depicts a scene from a recent stabbing incident on a London North Eastern Railway (LNER) train”, before giving otherwise correct details summarising the incident.

Separately, Google Lens’s AI overview linked a video of a confrontation on a train to the Huntingdon attack when we asked it about the clip, even though it is not real footage of the incident and was almost certainly generated with AI.

It comes after Full Fact revealed in August that Google Lens AI overviews had given users false and misleading information about images being shared widely on social media. We’ve also previously written about Grok misidentifying a viral video.

When we contacted Google about its AI overview responses this week, it told us that users can feed back on AI overview errors themselves, either by clicking a ‘thumbs down’ icon at the bottom of the AI overview box, or clicking the three-dot menu at the top-right of the box. We also contacted xAI for comment.

Are one in five children obese when they leave primary school?

New data obtained by Full Fact shows that official child obesity figures include tens of thousands of children who are not clinically obese, because they use a broader definition than doctors do.

This means the common claim that one in five children leave primary school with obesity – made by NHS England, and by health secretary Wes Streeting earlier this year, among many others – significantly overstates the proportion of children who are obese, if you define obesity clinically. In 2023/24 the proportion of children leaving primary school who are clinically obese was more like one in seven.

Childhood obesity is a very important subject, with big implications for the health and wealth of the whole country – no matter which threshold is used to measure it. But it’s vital that the statistics on childhood obesity used by politicians, government and the health service are clear and accurate.

The new data we obtained earlier this year through a freedom of information request to NHS England, which has since been published online, shows that in 2023/24 about 6% of 4-5-year-olds and 15.2% of 10-11-year-olds met the clinical thresholds for obesity. This compared to the 9.6% of 4-5-year-olds and 22.1% of 10-11-year-olds recorded as obese in the official NHS publication.

Full Fact’s analysis suggests this means 20,642 4-5-year-olds and 41,936 10-11-year-olds who are not clinically obese were counted as obese in the 2023/24 figures, which have since been used by ministers and the NHS.

The difference isn’t about how the children are measured. It’s the same height and weight data, gathered by the National Child Measurement Programme (NCMP). The distinction is which threshold is used to classify someone as obese.

Child obesity is assessed by comparing a child’s BMI to the range of measurements for their age and sex that was recorded between 1978 and 1994 (known as the UK90 reference data). Children with a BMI that puts them high on the scale are classified as overweight or obese.

In practice, to the nearest whole number, this means a child whose BMI puts them in the top 9% of the range for their age and sex in the UK90 data is considered clinically overweight by the NHS, and a child whose BMI is in the top 2% is considered clinically obese.

These are the thresholds that doctors and nurses are advised to use by the National Institute for Health and Care Excellence, and which are applied by the NHS’s own childhood BMI calculator. Parents are sent letters about the weight status of their children based on these thresholds too.

But these are not the thresholds that the NCMP uses in the data it publishes on overweight and obese children every year. It uses population monitoring thresholds instead, and categorises those in the top 15% of the range for their age and sex in the UK90 data as overweight, and those in the top 5% as obese – effectively including more children.

The Department of Health and Social Care told Full Fact that it and NHS England are transparent on the use of population monitoring and clinical thresholds, and how they differ. It said both are needed to assess children’s BMI, with population monitoring thresholds providing robust data on trends because they have been consistently used since the early 2000s.

The data we obtained under the FOI Act earlier this year was for 2023/24, but this week the government released new data for 2024/25. This included a more prominent explanation of the different categories, but the data itself did not show how many children belonged in each one.

That means we do not currently know how many children who are not clinically obese were included in the 10.5% of 4-5-year-olds and 22.2% of 10-11-year-olds counted as obese in the official statistics for 2024/25.

For more details, including more on why different definitions of child obesity are used, see Full Fact’s full investigation.