AKAnswered by Anna Kowalski · Last updated: November 2026
Short answer
Making an AI nude of a real person without their consent is unlawful in most of the developed world — criminal to share nearly everywhere, criminal to merely create in the UK and South Korea, civilly actionable in most US states. Fully synthetic nudes of invented adults occupy the lawful end. AI nudes of minors are prosecuted as child sexual abuse material in every jurisdiction — no exceptions, no "it's AI" defense. And "safe"? For the person uploading someone else's photo: the photo gets kept by an anonymous operator, and your payment trail exists. Not legal advice — for real situations, see a lawyer.
Is it legal to CREATE an AI nude?
Depends on who's depicted and where you are — the table is the honest answer:
Scenario
UK
US
EU
Korea
Synthetic adult, invented person
lawful
lawful
lawful + marking duties
lawful
Real adult, WITH verified consent
lawful
lawful
lawful
lawful
Real adult, NO consent — create only
criminal
criminal→civil→unaddressed, by state
criminalizing + GDPR violation
criminal
Real adult, NO consent — shared
criminal
criminal (federal + ~45 states)
criminal
criminal
Minor depicted — anything
CSAM felony
CSAM felony
CSAM felony
CSAM felony
The trend is one-directional: every year more jurisdictions move "create" into the criminal column. The UK's 2024 creation offence and Korea's production-and-possession regime are where the law is heading, not where it peaked.
What does US law say specifically?
Two layers. Federally, the Take It Down Act (May 2025) criminalizes knowingly publishing non-consensual intimate imagery — expressly including AI-generated depictions — with fines and up to two years (three where a minor is depicted, though minor cases route into harsher CSAM statutes), and requires covered platforms to remove flagged material within 48 hours of a victim request. Below that, roughly 45 states have their own statutes — most criminalizing distribution, a growing set criminalizing creation, and many adding civil causes of action with statutory damages so victims can sue without proving financial loss.
What about the UK, EU, and elsewhere?
UK: strictest mainstream regime — sharing criminalized under the Online Safety Act 2023; creating a sexual deepfake without consent criminalized in 2024. Creation alone is enough.
EU: directive 2024/1385 requires member states to criminalize non-consensual intimate deepfakes (transposition rolling through 2025–27); the AI Act adds mandatory marking; GDPR makes non-consensual processing of someone's likeness unlawful anyway.
South Korea: production AND possession criminalized, up to seven years — the post-2024 enforcement is aggressive and publicized.
Australia, Canada, most OECD: equivalent statutes in force or in late passage.
Is it safe — for the person doing it?
Set the law aside and the answer is still no, on four grounds the tools' own users concede:
Your upload is kept. Standard terms grant operators a perpetual license; uploads feed retraining corpora; several services' databases have already leaked — exposing victims' photos and users' payment details together.
Your payment is traceable. Card transactions, Telegram Stars, even "anonymous" crypto — all leave investigator-legible trails.
The service may be a scam. A meaningful share of listed sites deliver nothing, return generic output, or exist purely to harvest uploads.
The domain will die. Weeks-long lifespans mean credits and accounts evaporate routinely.
Is it harmful — for the person depicted?
Yes, and the mechanism is worth understanding: the harm doesn't depend on the image being real. An AI nude weaponizes a person's identity — their recognizable face in an intimate context they never chose — and that works on employers, family, and communities regardless of whether viewers could ever verify authenticity. The psychological literature treats synthetic-NCII victimization as equivalent to photographic image-based abuse; the threat "I have nudes of you" coerces identically whether the nudes were photographed or generated.
The asymmetry that defines the risk: the uploader risks prosecution and a data trail; the depicted person had no choice at all. Every safety framework in this space exists for the second person's benefit.
What can victims do — in order?
Preserve evidence. URLs, timestamps, screenshots — before anything gets removed.
Report via the platform's NCII channel. 48-hour removal duty for US-covered platforms under Take It Down.
Fingerprint the material. StopNCII (adults) / Take It Down (minors) hash-block re-uploads across participating platforms — without the image leaving your device.
De-index from search. Google and Bing run dedicated synthetic-NCII removal forms; search removal cuts exposure faster than chasing churned domains.
Police report + counsel. Post-2024 prosecutors bring these cases; civil statutes provide damages.
Get specialist support. Cyber Civil Rights Initiative (US), Revenge Porn Helpline (UK), national equivalents — synthetic victims are treated identically.