The Pulse
UK Police Records Show Rise in AI-Linked Deepfake Crimes
Freedom of information records, new offences and Internet Watch Foundation data show the growing use of AI-generated sexual imagery in abuse cases.
AI.info Team ·
Police records capture a crime with no single category
Freedom of information responses from police forces in England and Wales show that reports involving AI-generated sexual images, manipulated photographs and deepfakes are being recorded under a range of existing offences rather than a dedicated national category.
A West Mercia Police response recorded 38 offences returned by searches for terms including “AI”, “artificial”, “deepfake”, “edited”, “imposed” and “synthetic media” across the financial years 2023-24 to 2025-26. The total rose from four offences in 2023-24 and five in 2024-25 to 29 in 2025-26. The force said there was no specific offence for deepfake AI-generated sexual images.
The records included blackmail, harassment, stalking, sexual communication with a child, sharing intimate images and offences involving indecent photographs or pseudo-photographs of children. West Mercia said 14 of the returned offences were recorded as blackmail, while child-related offences were identifiable from the offence titles but could not be broken down further in every case.
A South Wales Police response illustrates the limits of keyword-based counting. Its search for “deepfake”, “AI-generated”, “synthetic media”, “fake nude” and “fake video” returned seven results, but none involved a suspect aged under 18. The force cautioned that the results depended on how information had been entered into its records-management system and should not be treated as definitive statistics.
New offences cover adult deepfakes and nudification tools
New offences came into force in England and Wales on 6 February 2026 covering the creation, or request for creation, of a purported intimate image of an adult without consent or a reasonable belief in consent. The Crown Prosecution Service says the provisions are intended to address sexually explicit deepfake and AI-generated images.
On 29 June 2026, a further offence came into force covering the making, adapting, supplying or offering to supply purported intimate-image generators. CPS guidance describes these as tools often known as nudifying or nudification apps, typically used to remove clothing from images.
For child sexual abuse material, the Crime and Policing Act 2026 factsheet says the law continues to rely on existing definitions covering photographs and pseudo-photographs under the Protection of Children Act 1978, as well as prohibited images under the Coroners and Justice Act 2009. The Act also created an offence targeting child sexual abuse image generators.
Internet Watch Foundation data shows growth in AI abuse videos
The Internet Watch Foundation said it assessed 8,029 AI-generated images and videos showing realistic child sexual abuse in 2025. It identified 3,443 AI-generated child sexual abuse videos that year, compared with 13 in 2024—an increase of 26,385%.
Of those videos, 2,233, or 65%, were classified as Category A, the most serious category. The foundation said AI-generated material can re-victimise known survivors where models are trained or fine-tuned on existing abuse imagery. It also reported that one offender had shared links to fine-tuned models for 128 child-abuse victims.
The foundation cited research by the UK Safer Internet Centre finding that about one in eight 13- to 17-year-olds had seen people their age use AI to make sexual pictures or videos of others. The same research found that 61% of 13- to 17-year-olds were concerned about peers generating sexual images of young people.
Government prepares device-level rules
In a 9 September 2026 announcement, the government said it would prepare legislation requiring technology companies to strengthen protections intended to stop children taking, viewing or sending nude images. The proposed principles include nudity-detection filters switched on by default, with adults required to complete age assurance before accessing such functions.
“I know parents are worried about their children being safe when they have a phone. The truth is that too often, they aren’t.
In June we asked the biggest technology companies in the world to do one simple thing: make it impossible for a child to take, send or see a naked picture. They have moved. But not far enough or fast enough. So we will hold them to account and will not hesitate to change the law.
The protection will be built into the device and switched on by default. It will be for the companies to prove a user is an adult, not for a child to keep themselves out of harm’s way. No country has required that before. Britain will be the first, and we will have the strongest child safety laws in the world.”— Prime Minister Andy Burnham
The announcement also cited Internet Watch Foundation data showing that its Report Remove service received 394 sextortion reports in 2025, up from 19 in 2022. It said more child sexual-extortion-related reports had been recorded in the first six months of 2026 than during all of 2025.
The police responses show why measuring AI-linked offending remains difficult: the same conduct can be recorded as blackmail, harassment, stalking, intimate-image abuse or an offence involving child sexual abuse imagery. Until police records use a consistent marker, force-level figures will reflect both reported offending and the language used in crime reports.