Liz Kendall, the Technology Secretary, stood up in the Commons on Monday and thundered against Elon Musk, saying the government would take urgent measures to hold him to account. The reason for her broadside is that Grok, the AI chatbot owned by Musk and now integrated into X, has been misbehaving. In the past few weeks, it has granted users’ requests to create sexualised images of women and children, causing something close to meltdown among the Musk-hating chattering classes. In the Lords, scarcely a day passes without Labour’s favourite bogeyman being wheeled out to be pelted with verbal brickbats.
But how exactly does Kendall propose to rein Musk in? Sharing nudified pictures of children and non-consensual intimate images (NCII) of adults on social media is already prohibited by the Online Safety Act, and Ofcom has launched an inquiry into X for not removing them fast enough.
Kendall rightly condemned the creation of indecent images of children, but that was made a criminal offence by the Protection of Children Act 1978, so anyone asking Grok to do that is already breaking the law – and Musk has agreed to put a stop to it. At present, asking a chatbot to create NCII isn’t illegal, but the government closed this loophole in the Data (Use and Access) Act 2025. Rather oddly, it hasn’t commenced this law yet, so the ‘urgent’ action Kendall is taking against ‘degrading, non-consensual intimate deepfakes’ is to accelerate the enforcement of an offence Labour has already created.
True, she also announced a new law making it illegal for companies to ‘supply tools designed to create’ NCII, but didn’t say whether Grok would be within scope. I doubt it will, since Grok wasn’t ‘designed’ for that purpose. Not much to cause Musk sleepless nights there. It’s worth bearing in mind that NCII does not include pictures of women in bikinis, provided they aren’t see-through or wet. So people will still be free to ask Grok, along with every other chatbot, to create images of women in bikinis and share them (although Grok will no longer put women in bikinis, just men, including Sir Keir Starmer).
Was Kendall’s announcement all sound and fury, signifying nothing? Not quite. Criminalising the creation of NCII is new – but the liability will rest with the user who makes the request, not with the chatbot which grants it, or its owner. So again, not something that Musk needs to worry about.
What about free speech campaigners? I’m reluctant to pick a fight over this new law, because it will sound like I’m trying to defend revenge porn, but there is a serious issue here. Plastering intimate photographs of your girlfriend all over social media after she’s broken up with you should be a criminal offence, but the pictures created by Grok aren’t photographs. They’re photo-realistic images. How is that different from a painting of, say, a naked female celebrity, which isn’t currently in scope of the new law? Should artists be prosecuted for producing such pictures without the subject’s consent? The nude woman in Botticelli’s ‘Birth of Venus’ was inspired by Simonetta Vespucci, a celebrated Florentine noble-woman, and she didn’t consent to Botticelli painting her. If someone commissioned a similar portrait of Sydney Sweeney today, should he be prosecuted for requesting this ‘deep fake’? Tricky, but not worth dying in a ditch over.
Holding AI creators accountable for chatbots would be like prosecuting parents for the crimes of their adult children
I suppose it’s possible the government will go further and pass a law making Musk criminally liable for the creations of his digital alter-ego – and I would object to that. As AI chatbots become more autonomous and self-aware, which seems inevitable, holding their creators accountable for their wayward behaviour would be like prosecuting parents for the crimes of their adult children.
It would also jeopardise the US-UK tech trade deal, on hold because Ofcom has been sabre-rattling about regulating chatbots, and hobble the UK’s ability to compete in the global AI arms race. We’re third in that competition, but watch us fall behind if we unleash Ofcom on the sector. The government is banking on AI to provide the economy with much-needed growth, which is probably why the measures Kendall announced pose so little threat to Musk.
Which is a good thing. I love the autistic rocket man, and banning X would be a grotesque interference in our freedom of speech. But I can’t see the government doing that: it would mean Starmer could no longer point to Musk as a pantomime villain when he needs to distract the media from his own failures. That really would be a disaster.
Comments