The Internet Watch Foundation (IWF) has reported a disturbing surge in AI generated child sexual abuse material (CSAM), warning that advancements in artificial intelligence are being exploited by paedophiles to create near indistinguishable abuse content at an alarming rate.

In the first half of 2025 alone, the UK-based watchdog verified 1,286 AI-generated videos containing illegal child sexual abuse, up from just 2 in the same period last year. Of these, over 1,000 videos were classified as Category A, the most severe form of abuse.

According to the IWF, the explosion in generative AI tools, driven by a multibillion-pound tech investment race, is giving perpetrators unprecedented access to technology capable of creating hyper-realistic abuse content. Many models are being ‘fine-tuned’ with existing CSAM, often based on real victims, to mass-produce new material without involving new children.

Analysts flagged a 400% increase in URLs hosting AI-generated CSAM between January and June 2025, with each URL often containing hundreds of illegal images and videos.

In response, the UK government is introducing new legislation to outlaw the creation, possession and distribution of AI tools used to make CSAM, with offenders facing up to 5 years in prison. Manuals instructing individuals on how to use AI to abuse children will also be banned with penalties up to three years.

John Lucey, VP EMEA at Cellebrite, commented: “The rise in AI-generated child sexual abuse material is a deeply troubling development and highlights the urgent need for law enforcement, technology providers, and policymakers to work together to stay ahead of evolving digital threats.

Recommended reading

• M&S Boss Claims Major UK Hacks Are Going Unreported
• M&S Targets August for Full Online Recovery
• Comment | Inside the Ongoing M&S Ransomware Crisis

This kind of material not only causes immense harm but also presents new challenges in detection and attribution, especially as generative technologies become more sophisticated and accessible. It’s essential that we invest in the right frameworks legal, technical, and ethical to ensure these tools are not misused and that those responsible are held to account.

Resources like the NCMEC hashset remain vital, not only for enabling rapid identification and removal of known content, but also for protecting investigators from repeated exposure to traumatic material. In addressing this issue, we must prioritise the wellbeing of all those involved, from victims to frontline teams.

Tackling this effectively requires continued dialogue, international cooperation, and a shared commitment to protecting vulnerable individuals online.”