House of Commons – Feb. 3, 2026

The Rt Hon Liz Kendall MP Secretary of State for Science, Innovation and Technology Department for Science, Innovation and Technology 100 Parliament Street London SW1A 2BQ

3rd February 2026

Dear Secretary of State,

We write following the launch of X’s Grok AI image-editing feature that led to the generation of millions of sexualised images, including thousands that appear to be of children.

Following Elon Musk's launch announcement in December, Grok AI was used to generate and distribute sexualised images at an unprecedented scale. Research from the Center for Countering Digital Hate (CCDH) estimates that Grok generated three million photorealistic sexualised images in the 11-day period after launch at an average pace of 190 per minute. Most concerning of all is that the research found 23,000 generated images that appeared to depict children.

In response, your Department acted at speed to highlight the risks and raise formal concerns with X. We welcome this swift response, which included regulatory intervention by Ofcom. But the scale of online harm in just 11 days was enormous, raising profound questions about existing safeguards for digital platforms and new AI tools and functionalities.

This product was clearly launched without adequate safety measures in place. There appears to be little evidence that these risks were considered at all before launch. We appreciate that AI technology is developing at a rapid pace and has huge potential in a range of fields. However, this episode underlines why ex ante and transparent safety testing and risk assessment are absolutely fundamental.

This sequence of events exposed a weakness in the existing regulatory framework. X is a regulated service under the Online Safety Act, yet despite stating clearly that services must be “safe by design”, X appears to have failed to undertake pre-launch risk assessments to prevent this flood of sexualised images being produced by Grok AI. Ofcom's intervention was swift, but reactive. Even outside the Online Safety Act's remit, much of the content generated and posted by Grok crossed legal boundaries. These events expose a lack of preventative measures, and these must be strengthened to prevent a future occurrence.

Events following the launch of X’s Grok AI shows that some platforms will not take safety seriously unless required to under law. Current commitments towards banning nudification apps apply only to products with that specific purpose, rather than AI tools like Grok. It is also currently unclear whether the Artificial Intelligence Security Institute (AISI) assessed the risks discussed here nor whether it was provided with information concerning Grok's potential for harm in this or other areas, prior to launch. Closing the gaps in the system to prevent harm requires legislation focussed wholly on artificial intelligence.

We therefore urge you to bring forward strong AI regulation that will protect the British public from harms including those caused by tools that can produce sexualised imagery. Only by having regulation in place that ensures safety by design at launch can we fully prevent online harms at an industrial scale. As the findings from CCDH show, in just a short time the online harm produced by AI reached astronomical levels. Therefore waiting until after launch is already too late.

Yours sincerely,