Leadership1 publisher2 min readPublished
IWF's half-year count of AI child abuse images passes 2025's full-year total by 40%
Internet Watch Foundation analysts assessed 6,310 AI-generated child abuse images in the first half of 2026, 40% more than in all of 2025. The charity wants binding AI law, but the government has not signalled an imminent bill, so existing criminal offences remain the rules image firms face.
The Board Room · Leadership desk
Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction
What happened
- The IWF figures cover still images only, although AI-made abuse videos are where the surge has been sharpest.
- Most of the images analysts reviewed depicted girls, and most depicted children aged between seven and 13.
- Report Remove has had 420 reports this year from children who believed images of themselves were faked to look explicit, already above its 2025 total of 397.
- The National Crime Agency and the IWF are advising parents to keep children's pictures off social media, or share them only with a close-friends group.
Compiled by The Board RoomSomething wrong?How this is made
Why it matters
- exposure Developers already sit inside UK criminal law: building a model meant to produce hyper-realistic abuse exposes its maker to up to five years, whoever types the prompt.
- precedent If parliamentarians' call for binding legislation succeeds, duties would attach to how every image model is built, a wider scope than offences tied to abusive design or adaptation.
- decision With no imminent bill, how far to restrict an image generator stays each company's own call this quarter, and the IWF's next count will bear on how long that discretion lasts.
The 40% comparison sets six months of 2026 against all twelve of 2025 [2]. If the second half of this year only matched the first, the IWF's count would reach 12,620 images, about 2.8 times last year's total of more than 4,500 [1]. The projection assumes a steady pace. It also inherits the count's limits, because the tally leaves out video [3].
The government says the law already covers this. "UK law is clear that child sexual abuse material is illegal, regardless of whether it is AI-generated or not," a government spokesperson said [12]. "We are going further by banning AI models that are used to create this type of abuse material," the spokesperson added [12]. It is illegal to adapt an AI model to create abuse material or to distribute such a model [10]. Developing a model designed to produce hyper-realistic abuse material carries a sentence of up to five years [11].
Those offences turn on what a model was designed or adapted to do. The IWF said its figures underline the need for new statutory regulation on AI [5], and its ask goes further than intent. Hannah Swirsky, its head of policy, backed calls from parliamentarians for "binding legislation on AI" to compel companies to build safer models [8]. "It is incumbent on tech companies to build tools which cannot be abused this way," she said [9]. In my view the two positions differ on who carries the obligation. On the offences as reported, a general-purpose image generator built in good faith appears to fall outside the design-for-abuse provisions [10][11]. Swirsky's standard would bring it in [9].
On timing, the record points away from this quarter. Andy Burnham's government has not signalled that a bill focused on AI is imminent [6]. For a company that hosts or generates images, the rules in force now are the criminal offences above. The evidence shows pressure building for tighter rules. It does not show a timetable. The bar a future bill must clear comes from one minister. Kanishka Narayan, the AI minister, has said "nothing is off the table" on fresh regulation, and that the test for any proposal will be: "Will it make the British people safer?" [7]
What to watch
- Whether the government signals an AI bill, and whether it puts design duties on general-purpose model builders or only extends the existing offences.
- The IWF's full-year 2026 figures, and whether it publishes a count for AI-generated video alongside still images.
- How Kanishka Narayan's 'safer' test is applied to the binding-legislation proposals parliamentarians are backing.