General Discussion

sl8

(16,947 posts) Wed Nov 5, 2025, 03:06 PM Nov 5

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

https://fullfact.org/crime/grok-google-lens-ai-imagery-train-attack/

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

5 November 2025

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

Grok, the AI chatbot created by Elon Musk’s start-up xAI, told users one AI-generated image of an injured man lying in a carriage was not AI-generated, and said it “appears to be a genuine photo”. 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 an article from the organisation about the Huntingdon attack.

Google Lens’s AI overview also linked a video of a confrontation on a train carriage to Saturday’s attack, 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.

[...]

2 replies

= new reply since forum marked as read

Highlight:

Grok and Google Lens AI overviews claim fake imagery shows Huntingdon train attack (Original Post) sl8 Nov 5 OP

If there's one thing you REALLY shouldn't trust free AI tools for it's to validate that photos and videos are legitimate AZJonnie Nov 5 #1

Of the 2 photos at the link, the one on the right is laughably wrong muriel_volestrangler Nov 5 #2

AZJonnie

(2,249 posts)

1. If there's one thing you REALLY shouldn't trust free AI tools for it's to validate that photos and videos are legitimate

Reply to sl8 (Original post)

Wed Nov 5, 2025, 04:19 PM

Nov 5

In no small part, because some people who make fake images also know how to trick AI's into being more likely to believe they are real images, and thus tell users that they are.

I think it's safe to predict there will an "arms race" in this regard, similar in concept as the one between the anti-virus companies, and the virus-makers.

muriel_volestrangler

(105,192 posts)

2. Of the 2 photos at the link, the one on the right is laughably wrong

Reply to sl8 (Original post)

Wed Nov 5, 2025, 05:05 PM

Nov 5

The attack happened after sunset, in a train that was then stopped in a station in a town. It shows paramedics treating a casualty in a train in daylight, with countryside outside the window. With a suspect being led away, past loads of people still in their seats. Whenever it was produced, it took little time for the 2 people initially arrest to be identified as Black, and the suspect is white; and the first reports were all about how everyone had run from the attacker. The person who chose that picture wasn't even trying to make it believable.

Reply to this discussion