TL;DR
The Internet Watch Foundation (IWF) found more photorealistic AI-generated child sexual abuse imagery in the first six months of 2026 than in the whole of 2025, the Guardian reported on Monday 5 October. Its analysts assessed 6,310 such images that meet the legal definition, 40% above last year’s total. The UK charity says the numbers make the case for statutory regulation of AI.
What the figures show
Last year’s total was a little over 4,500 images. The new count covers still images only, so it leaves out AI-made video, where the growth has been steepest. Most of the images showed girls, and most showed children between seven and 13.
A second measure points the same way. Report Remove, which helps young people stop intimate pictures of themselves spreading online, has logged 420 reports in 2026 from children who believed images of them had been faked or altered to look explicit. Its figure for all of 2025 was 397.
The push for legislation
Hannah Swirsky, the IWF’s head of policy, backed parliamentarians calling for “binding legislation on AI” that would force companies to make safer models. “It is incumbent on tech companies to build tools which cannot be abused this way,” she said.
The government has not indicated that an AI bill is close. Kanishka Narayan, the AI minister, has said “nothing is off the table”, but that any proposal will be judged on one question: “Will it make the British people safer?” Last week he told regulation advocates to show their working at Labour’s conference.
Existing law covers some of this. AI-generated abuse imagery is already a crime in Britain, as is adapting a model to produce it or passing such a model on. Building a model designed to create hyper-realistic abuse material carries a maximum five-year sentence. Ministers are, in a spokesperson’s words, “going further by banning AI models that are used to create this type of abuse material”.
Advice for parents
The National Crime Agency and the IWF are urging parents and guardians not to post children’s photos on social media, or to share them only from private accounts or with a “close friends” group. That echoes their May warning to schools about pupil photos being manipulated for extortion.
Looking forward
The IWF’s argument shifts the focus from offenders to model builders. In our view, the test Narayan has set gives the charity an opening: its figures are a direct measure of harm to British children, and the second-half count for 2026 will show whether the trend is still climbing. For UK firms that build on or fine-tune image models, the existing offences around adapting and distributing models are already a compliance point, whatever happens to a new bill.