UK law under the Protection of Children Act 1978 already makes creating, possessing or distributing AI-generated CSAM illegal with the same penalties as material involving real children. The Crime and Policing Act 2026 adds new offences for CSA image-generators and AI-optimised models with maximum sentences of five years, while updating paedophile manual rules to cover AI guidance.
The distinction between synthetic and human-made material was treated here as a loophole waiting to be exploited, with Yes winning by nine to one. This was a self-selected online audience rather than a national poll, but its unanimity was striking.
The floor’s most-respected Yes case warned that leniency would create “a disastrous loophole”, allowing offenders to reach for an “I thought it was AI” defence. Another highly rated voice put the enforcement problem more starkly: “I wonder if you could tell the difference?”
No voter left an argument on the page, so the opposing case — that AI material involves no directly abused child and might warrant a different penalty — went unanswered rather than defeated in debate. The nearest concession came from a Yes voter who acknowledged that “no victim is harmed in the AI version BUT” immediately returned to the risk of normalisation.
The discussion therefore centred less on competing theories of harm than on whether the law could afford a distinction in practice. “CSAM is CSAM whether AI or not” captured the chamber’s governing instinct, while the less-respected contributions pushed beyond equivalence towards still harsher punishment and furious denunciations of institutional failure.
The verdict was not just for equal penalties, but against leaving offenders or investigators any ambiguity to work with.
Should the UK replace human rights treaties with its own bill of rights?
Justice · 497 votes · 61% voted Yes
Are prisoner rehabilitation programmes worth the time, money, and effort?
Justice · 453 votes · 62% voted Yes
Should the UK legalise approved pepper spray for adults to carry with training for self-defence?
Justice · 453 votes · 56% voted Yes
Should the UK end Sharia Councils?
Justice · 440 votes · 72% voted Yes
Should non-crime hate incidents be scrapped?
Justice · 440 votes · 78% voted Yes
Should litterers receive community service picking litter in addition to a fine?
Justice · 393 votes · 88% voted Yes
Should Palestine Action remain proscribed as a terrorist organisation?
Justice · 393 votes · 68% voted Yes
Should convicted offenders have to repay the cost of their punishment?
Justice · 323 votes · 52% voted No
Every vote you cast builds your political profile — where you lean, worked out from real questions rather than fixed statements.
10 Opinions
Making AI CSAM less harshly punished than offences involving real children normalises child abuse and would lead to it being harder to punish offenders due to an ‘I thought it was AI’ defence
One thousand percent yes. Otherwise "I thought it was AI" will be the go to defence for offenders, and it will be very difficult to prove otherwise.
CSAM is CSAM whether AI or not. If someone is convicted for having CSAM, the police investigating should not be expected to determine if every image is real or generated. They already have a hard enough time having to see the images and videos as it is.
Check the hard drives of anyone who votes no
I wonder if you could tell the difference? Do not leave gaps where evil can slip through
Yes. CSAM is disgusting, even if it is just a facsimile of a child. Same with things like Lolicon. Ban the lot.
Nonces need the woodchipper
If anything the penalties should be far higher. OK, I understand that no victim is harmed in the AI version BUT it is aimed at a specific audience, gratifying and perpetuating their sick needs and to some extent normalising it for them…
We are way too soft on child abuse. Paedophilia is a disease cured only by death. The real worry is enablers who need life in prison. My concern is this evil is everywhere the authorities are doing nothing! Why? Saville= Not one enabler arrested the BBC and St James hospital had many. Rochdale and others= no enablers arrested Torbet and 8 others= Scottish govt cover up enablers protected. Bennell= Not one enabler arrested his case is linked to Torbet case
Treating AI gen'd material more leniently would create a disastrous loophole, encouraging offenders to claim genuine abuse footage is merely AI-generated to evade justice. AI creations of this sort will still fuel grooming, and normalise abuse