Musk’s Grok AI bot falsely suggests police misrepresented footage of far-right rally in London
https://www.theguardian.com/technology/2025/sep/14/musks-grok-ai-bot-falsely-suggests-met-misrepresented-footage-of-clashes-with-far-right
Posted by FeigenbaumC
7 Comments
It’s common knowledge Grok just parrots far right talking points, right? Every time it accidentally tells the truth Musk puts in more safeguards to censor it again
No one should take it seriously
Yeah. I managed to convince Grok that women could carry children and give birth through their asshole. Grok came up with insanity that these children would be small, grotesque and mutilated and the mothers likely wouldn’t survive the birth.
So yeah. Grok is fun and Grok will say anything you want it to say.
Is this the A.I. that refers to itself as *’MechaHitler’*?
>Grok, which has had a track record of giving false and misleading answers, replied: “This footage appears to be from an anti-lockdown protest in London’s Trafalgar Square on 26 September 2020, during clashes between demonstrators and police over Covid restrictions.”
At least it said ‘appears to be’ rather than confidently stating it as fact.
Now look up how often google AI gets things badly – and confidently – wrong. You can often get completely contradictory answers from google AI just by slightly rewording it (but with the same meaning).
So it could be that Musk has told Grok to post this particular conspiracy theory or it could just be that AI in general is still shite and not reliable.
Grok is being sent for re-education.
https://i.imgur.com/fGPfBdx.jpeg
Grok has been re-educated.
https://i.imgur.com/mXkgukp.jpeg
Stop. Using. Twitter.
This is the message that you should be constantly telling your family, friends, politicians, and organisations.
This is part of the problem with AI. It is being trained using a variety of internet sources, many of which are wrong to varying extents. Grok in particular is terrible based on what nonsense gets shared on Twitter. It gets worse because people cite what Grok says in their comments and this gets fed back into Grok for further education. We have AI being trained using AI-generated lies.