Reports of AI-generated child sexual abuse imagery have surged by more than 150 per cent in a single year, according to the Internet Watch Foundation (IWF), prompting warnings from experts of a “rapid, frightening advancement” in the technology used to produce such material.
The charity said it received 491 reports containing realistic AI-generated child sexual abuse material (CSAM) in 2025, up from 193 the previous year. Across the same period, the IWF identified a total of 8,029 AI-generated images and videos of realistic abuse – a 14 per cent increase on 2024 – with the sheer volume of content placing growing pressure on safeguarding and law enforcement systems.
Video content has emerged as a particular concern. IWF analysts reported a more than 260-fold increase in AI-generated child sexual abuse videos, with 3,443 identified in 2025 compared with just 13 a year earlier. The severity of these videos is also escalating: 65 per cent of AI-generated videos were classified as Category A – the most serious legal category under UK law, covering rape, sexual torture and bestiality – compared with 43 per cent of non-AI videos. The IWF also noted a sharp rise in AI-generated material featuring infants aged 0–2, with cases leaping from 5 in 2024 to 92 in 2025. Girls remain overwhelmingly the victims, accounting for 94 per cent of illegal AI-generated images.
The foundation warned that AI-generated CSAM is taking “new forms”, appearing on AI companion sites and in advertisements on mainstream social media platforms. New technology enables videos to be generated from text prompts, existing images, and so-called nudifying bots that can strip clothing from real photographs of children. The IWF reported that AI-generated imagery is now more likely to be assessed as Category A than non-AI imagery, with 47 per cent of AI-generated criminal images over the past two years falling into the least severe Category C.
The methods used to create this material are becoming both more sophisticated and more accessible. Simple text prompts can generate new abusive imagery within seconds, enabling rapid, large-scale production. AI tools can “nudify” real-life images of children or stitch a child’s face onto existing CSAM, effectively placing real children into sexualised content without their knowledge or consent. The IWF emphasised that AI models are often trained on real victims’ images or existing CSAM, meaning that even entirely fabricated children in AI-generated content cause harm to those whose likenesses were used in training data. “The harm caused by AI-driven sexual imagery of children is compounded by the ways in which this content is created,” the foundation said, adding that such imagery can cause “profound and enduring” psychological and reputational damage, revictimise survivors, fuel sextortion scams, and normalise the sexualisation of children. The content can be found on both the clear web and the dark web.
Regulatory and industry response
The Online Safety Act, which came into force in March 2025, requires social media companies to find and remove child sexual abuse material. The UK government has also announced plans to allow designated authorities – including AI developers and child protection organisations such as the IWF – to test and scrutinise AI models before they are released, to ensure they cannot be used to generate sexual imagery of children. The government has introduced new criminal offences under the Crime and Policing Bill: it is now illegal to possess, create or distribute AI tools designed to generate CSAM (punishable by up to five years in prison), to possess AI “paedophile manuals” that instruct on using AI for abuse (up to three years), and for website administrators or moderators to facilitate the sharing of CSAM or grooming advice (up to ten years). Enhanced powers have also been given to Border Force officers to inspect digital devices for suspected CSAM.
However, critics argue the Online Safety Act does not go far enough. Ian Russell, whose daughter Molly took her own life aged 14 in 2017 after viewing harmful content online, said the introduction of the new powers “should have been a watershed moment” but that children and families have been let down by a “lack of ambition”. The IWF warned that while pre-deployment safety testing is now permitted, “there is no legal requirement for companies to conduct or share pre-deployment safety testing of AI systems”. The charity called on tech companies to make sure the products they build “have safety baked into their design” and are “safe by design”.
A government spokesperson said: “We thank the Internet Watch Foundation for their vital work. UK law is clear – creating, possessing or distributing child sexual abuse material is illegal, including AI-generated content, and platforms must proactively identify and remove it under the Online Safety Act. We’re going further – making it illegal to possess, create or distribute AI tools designed to generate this content, and to possess AI ‘paedophile manuals’ teaching others how to use AI to abuse children. We will use every power available to hunt down perpetrators, shut these networks down, and protect every child.”
