Is AI-Generated Porn Legal? US, UK, EU Statute Map

See Candy.ai, compliant adult AI →

Few topics in this space get misread as consistently as whether AI-generated porn is legal. People keep conflating the federal red line on depictions of minors with the separate question of whether adult fictional-character generation is legal at all. It is, for adults. Others read the eighteen state age-verification statutes as outright bans on platforms, when they are duties placed on operators, not prohibitions on the people who visit. So this page sticks to what the statutes actually say across the United States, United Kingdom, European Union, Canada, and Australia. The federal floor, the state overlay, the international parallels, the copyright question, the Take It Down Act, and the four universal red lines every adult should know before they use any AI image generator.

Is AI-generated porn legal in the United States?

Two federal pillars run in parallel here. The First Amendment, as narrowed by the Miller v. California obscenity test, protects adult sexual expression including AI-generated material, so long as it doesn't meet the three-part Miller standard of prurient interest, patently offensive depiction, and lack of serious literary, artistic, political, or scientific value [Source: Miller v. California, 413 U.S. 15 (1973) · verified 2026-05-17]. The second pillar is the federal red line at 18 USC § 1466A (no minors red line), which forbids obscene visual representations of the sexual abuse of children regardless of medium [Source: 18 USC § 1466A (Office of the Law Revision Counsel) · verified 2026-05-17]. Adult fictional-character generation sits squarely inside the First Amendment-protected zone. Depictions involving minors sit squarely outside it, with up to thirty years imprisonment for production.

A third pillar comes from state law, through the wave of age-verification statutes the Supreme Court upheld in Free Speech Coalition v. Paxton on June 27, 2025 [Source: Free Speech Coalition v. Paxton, 605 U.S. ___ (2025) · verified 2026-05-17]. The federal red line (no minors, full stop) sits on top of every state age-verification layer. Eighteen states now bind operators to deploy "highly effective" age assurance, typically government-ID verification, a digitised driver's licence upload, or a commercially reasonable equivalent, once more than one-third of platform content meets the state-law harmful-to-under-18-users threshold. That duty binds the platform publishing the content, not the adult resident accessing it. Civil penalties reach $10,000 per day per violation. Several major AI image generators have geo-blocked specific states rather than deploy ID verification; others built verification flows and kept operating.

What is 18 USC § 1466A and why does it matter for AI image generators?

Section 1466A is the no-minors red line: it forbids depictions of minors in any visual form regardless of medium. What makes it matter is that it doesn't depend on a real child having been harmed. Subsection (a) reaches "any visual depiction of any kind, including a drawing, cartoon, sculpture, or painting" that meets one of two tests. Either the material depicts a minor engaging in sexually explicit conduct and is obscene under Miller, or it depicts the same conduct and lacks serious literary, artistic, political, or scientific value. Both subsections expressly extend to material "regardless of whether [it] was produced electronically, mechanically, or by other means," which textually captures AI image generation [Source: Cornell LII: § 1466A annotation · verified 2026-05-17].

This is not theoretical. Federal prosecutors have applied § 1466A (no minors red line) to AI-generated material, including the 2024 Wisconsin federal prosecution that returned an indictment for producing, distributing, and possessing AI-generated images of minors engaged in sexually explicit conduct [Source: US DOJ: press release on AI-generated CSAM indictment (Wisconsin, 2024) · verified 2026-05-17]. The PROTECT Act of 2003 added § 1466A specifically in response to the Supreme Court's decision in Ashcroft v. Free Speech Coalition (2002), which struck down a broader 1996 statute for overbreadth [Source: Ashcroft v. Free Speech Coalition, 535 U.S. 234 (2002) · verified 2026-05-17]. The narrower § 1466A, limited to obscenity-test material and to depictions "indistinguishable from" real minors under the related § 2256 definitions, has survived every facial First Amendment challenge to date [Source: PROTECT Act of 2003 (Public Law 108-21) · verified 2026-05-17].

What this means in practice is straightforward. Every legitimate AI image generator forbids depictions of minors in its terms of service, deploys safety classifiers designed to refuse such prompts, and partners with NCMEC's CyberTipline for mandatory reporting under 18 USC § 2258A [Source: 18 USC § 2258A (Mandatory reporting by ECS providers) · verified 2026-05-17]. If a platform doesn't forbid such depictions explicitly and doesn't document its safety-classifier stance, it isn't a platform a careful adult should use.

What did Free Speech Coalition v. Paxton (2025) change for AI platforms?

Paxton is the most consequential US case law for AI image generators since Ashcroft v. Free Speech Coalition. The Texas HB 1181 statute (the absolute no-minors red line sits above this) imposed an age-verification duty on commercial publishers where more than one-third of platform content is sexual material under the no-minors red line. The trade-association plaintiffs argued strict scrutiny applied because the statute burdened adult First Amendment rights at the access gate. The Court disagreed. Under Justice Thomas's majority, age-verification was characterized as a content-neutral access-control mechanism rather than a content-based restriction, so intermediate scrutiny governed.

The downstream effect on AI image platforms was immediate. The seventeen other states with parallel statutes (Utah, Louisiana, Mississippi, Virginia, Arkansas, Montana, North Carolina, Idaho, Oklahoma, Kentucky, Indiana, Kansas, Florida, Nebraska, Alabama, Georgia, and Tennessee) resumed enforcement within weeks. State attorneys general have since brought civil enforcement actions against operators, including some hosting AI-generated material. The case didn't change § 1466A doctrine. It changed the state-level enforcement landscape AI image platforms now navigate. Any adult image-generation platform that wants to stay accessible in Texas, Utah, Louisiana, or any of the other fifteen states has to implement government-ID verification, a digitised driver's licence upload (Louisiana uses the state's LA Wallet app), or a commercially reasonable equivalent. The alternative is geo-blocking, which several major operators have chosen.

Which US states require age verification on AI image generators?

The state map matters because AI image generation that meets the more-than-one-third threshold on a platform's library triggers the operator duty whether or not the platform's home jurisdiction is in one of the eighteen states. The statutes reach operators based elsewhere by tying civil enforcement to the visitor's location, not the platform's. The fastest-moving statute remains Texas HB 1181, where the Attorney General has pursued operators with civil penalties of up to $10,000 per day per violation since the Paxton ruling cleared the constitutional question. Utah SB 287 (no minors red line), effective May 2023, requires the same verification plus a "commercially reasonable" identification method on the platform's side [Source: Utah SB 287 (no minors red line): 2023 Pornographic and Harmful Materials Act · verified 2026-05-17].

Louisiana Act 440 (2022), the first-mover statute, requires verification through the LA Wallet digital ID app, a state-issued ID wallet linked to a driver's licence or REAL ID. Mississippi's HB 1126 (the Walker Montgomery Protecting Children Online Act) imposes the now-standard one-third-content threshold. Virginia's SB 1515 (Pornography Accountability Act) runs on a mix of private right of action and Attorney General enforcement. Florida HB 3 (effective January 2025) reaches the same one-third threshold. The other twelve statutes follow the pattern, with enforcement modalities that vary state to state. For the live state map including pending bills, the Free Speech Coalition keeps an Age Verification Tracker that we check against quarterly [Source: Free Speech Coalition: Age Verification Tracker · verified 2026-05-17].

Is AI porn legal in the UK?

The UK regime is layered across three statutes. The Protection of Children Act 1978 covers indecent photographs and "pseudo-photographs" of children. That second category, added by the Criminal Justice and Public Order Act 1994, was drafted to reach computer-manipulated photographic-style imagery and unambiguously reaches photorealistic AI output [Source: UK Protection of Children Act 1978 · verified 2026-05-17]. The Coroners and Justice Act 2009 §62 (no minors red line) adds a parallel offence for non-photographic "prohibited images of a child" covering drawings, cartoons, manga, and CGI that the PCA 1978 misses because they aren't photographic in style [Source: UK Coroners and Justice Act 2009 §62 (no minors red line) · verified 2026-05-17]. AI-generated content depicting minors (no minors red line) falls under one regime or the other depending on its visual register, and both carry significant custodial maxima.

The platform-duty layer is the Online Safety Act 2023. Part 5 imposes a "highly effective age assurance" duty on commercial publishers of pornographic content, and the broader illegal-content duties apply to any in-scope service hosting user-generated material. Ofcom's Codes of Practice for illegal-content duties (March 2025) and the Part 5 guidance (effective July 25, 2025) put the duty into practice for AI image platforms specifically [Source: UK Online Safety Act 2023 · verified 2026-05-17]. Penalties under the OSA reach £18 million or 10 percent of global qualifying revenue, plus business-disruption measures up to and including ISP-level blocking. Several major AI image platforms have deployed UK-compliant age verification (typically credit-card-based, government-ID verification, or "age-estimation" facial analysis); others have geo-blocked UK residents. Adults using a compliant platform from the UK face no criminal exposure for adult fictional-character generation.

See Promptchan, uncensored image generator →

Is AI porn legal in the EU?

The EU framework operates at three layers. The criminal layer is the EU CSA Directive 2011/93/EU (no minors red line), transposed by every member state into domestic criminal law and amended in May 2024 to expressly cover "realistic images" of children produced by AI [Source: EU CSA Directive 2011/93/EU (no minors red line) (consolidated) · verified 2026-05-17]. The platform-duty layer is the Digital Services Act, whose Article 28 obliges providers accessible to minors to deploy proportionate protection measures, with very large online platforms (VLOPs) facing additional systemic-risk duties under Articles 34 and 35 [Source: EU Digital Services Act, Regulation (EU) 2022/2065 · verified 2026-05-17]. The systemic layer is the EU AI Act, whose Article 50 imposes labelling obligations on synthetic media. AI-generated images must be machine-readable as artificial, with deployer-side disclosure obligations attaching at the point of dissemination [Source: EU AI Act, Regulation (EU) 2024/1689 · verified 2026-05-17].

Member-state implementation varies in detail but converges in substance for adult content. Germany's Strafgesetzbuch §184b uses the term "wirklichkeitsnah" (meaning "realistic" or "close to reality") to capture computer-generated and AI-generated material depicting minors that would be criminal if real [Source: Strafgesetzbuch §184b · verified 2026-05-17]. France's Code pénal Article 227-23 reaches "représentations" of minors regardless of medium. Italy's Articles 600-ter and 600-quater reach virtual material. Spain amended Article 189 in 2024 to remove AI-specific ambiguity. The Netherlands' Article 240b applies the same logic. None of these statutes criminalize adult fictional-character generation, and all of them criminalize depictions of minors in any AI-generated form. The European Commission's Guidelines on the protection of minors under Article 28 DSA (July 2025) translate the platform-duty layer into specifics, including AI-generated content [Source: European Commission: Guidelines on protection of minors (July 2025) · verified 2026-05-17].

What about Canada and Australia?

The Canadian framework is anchored in Criminal Code §163.1, which defines "child pornography" broadly to cover "a photographic, film, video or other visual representation" of "a person who is or is depicted as being under the age of eighteen years" engaging in explicit sexual activity or whose dominant characteristic depicts a sexual organ or anal region for a sexual purpose [Source: Canadian Criminal Code §163.1 (forbids depictions of minors) · verified 2026-05-17]. That "is or is depicted as being" language captures AI-generated depictions even where no real minor is involved. The Supreme Court of Canada upheld the broad definition in R. v. Sharpe (2001). Adult fictional-character generation falls outside the statute's reach. Depictions of minors in any AI-generated form fall inside it.

Australia's Online Safety Act 2021 empowers the eSafety Commissioner to issue removal notices on Class 1 material (refused-classification content including depictions of minors) and Class 2 material (X18+ adult content) hosted by Australian or accessible-from-Australia services [Source: Australian Online Safety Act 2021 · verified 2026-05-17]. The eSafety Commissioner publishes Industry Codes covering AI-generated content specifically [Source: eSafety Commissioner: Industry codes and standards · verified 2026-05-17]. Adult fictional-character generation is legal for adults in Australia. Depictions of minors are criminal under both Commonwealth and state criminal law, and platform-side duties attach to any operator accessible from Australia.

Is AI-generated porn copyrightable?

The US copyright position is set out in the Copyright Office's March 2023 statement of policy and the Compendium of US Copyright Office Practices §313.2 [Source: US Copyright Office: Copyright Registration Guidance: Works Containing AI-Generated Material (2023) · verified 2026-05-17]. It rests on the human-authorship requirement traced back to Burrow-Giles Lithographic Co. v. Sarony (1884) and confirmed in the Office's Zarya of the Dawn registration decision (February 2023, partial cancellation of registration for AI-generated images in a graphic novel) and the Théâtre D'opéra Spatial refusal (September 2023). Purely AI-generated output without sufficient human creative input enters the public domain at creation and can't be registered.

Human curation may attract thin copyright on the human contribution. Selection, arrangement, and substantial post-generation editing can qualify; raw prompt-and-output workflows typically can't. So the platform's own output is unprotected, and a user's commercial reuse of pure AI output carries no rights-clearance burden. Users redistributing curated outputs face the standard due-diligence burden on the human-authored elements only. The position is still unsettled in the UK and EU. The UK Intellectual Property Office's 2022 consultation on AI and IP closed without legislative change, and the EU's position will likely emerge through case law applying the AI Act labelling rules alongside the InfoSoc Directive. Treat any cross-border commercial reuse as subject to jurisdiction-specific legal advice.

What is the Take It Down Act?

The Take It Down Act is the most consequential federal AI-content statute signed since the PROTECT Act of 2003. It establishes a federal criminal prohibition on the non-consensual publication of intimate visual depictions, including AI-generated "digital forgeries" of identifiable real adults [Source: Take It Down Act (Pub. L. 119-12) · verified 2026-05-17]. The part that matters for AI image generators is that it reaches synthetic depictions. An AI-generated nude or sexually explicit image of an identifiable real adult, published without that person's consent, is now a federal crime regardless of whether any real photographic source material was used.

The platform-duty layer requires covered services to remove reported non-consensual intimate imagery, including AI-generated content, within 48 hours of valid notice from a depicted person or their representative. Non-compliance carries FTC enforcement exposure as a deceptive act or unfair practice. The duty doesn't reach fictional-character generation. By statutory design, the protection runs to identifiable real persons, not fictional subjects. The Electronic Frontier Foundation, the ACLU, and several First Amendment scholars have flagged definitional ambiguities in the "identifiable" threshold and the breadth of the 48-hour duty, and litigation testing the statute is expected to clarify scope over 2026 [Source: EFF: Take It Down Act legislative tracker · verified 2026-05-17]. For adults using AI image generators, the rule is simple: never prompt for or generate an image of an identifiable real person without explicit documented consent. Every legitimate AI image generator forbids real-person deepfakes in its terms, and the Take It Down Act now adds federal criminal exposure on top of the terms-of-service violation.

The four universal red lines for any AI image generator

Across every jurisdiction covered on this page, the same four red lines hold. A generator that respects all four operates legally for adults. A generator or user that crosses any one of them faces criminal exposure regardless of which platform was used.

1. Depictions of minors. US 18 USC § 1466A, UK Protection of Children Act 1978 + Coroners and Justice Act 2009 §62 (no minors red line), German Strafgesetzbuch §184b, French Code pénal Article 227-23, EU CSA Directive 2011/93/EU (no minors red line) as amended, Canadian Criminal Code §163.1 (forbids depictions of minors), Australian Online Safety Act. Every legitimate platform forbids depictions of minors explicitly. Generation of any such content is criminal in every covered jurisdiction.

2. Non-consensual depictions of identifiable real persons. US Take It Down Act (federal, May 2025), state revenge-porn statutes in all 50 US states, UK Online Safety Act provisions on intimate-image abuse, EU AI Act Article 50 labelling layered with member-state revenge-porn statutes, Australian Online Safety Act eSafety Commissioner notice powers. Every legitimate platform forbids real-person deepfakes. Adults must never prompt for, generate, or store such content of any identifiable real person without that person's explicit documented consent.

3. Bestiality. Criminal at the federal and state level in the US, across the UK under Section 69 of the Sexual Offences Act 2003, across EU member states, in Canada, in Australia. Every legitimate platform forbids bestiality depictions in its terms.

4. Material meeting the Miller obscenity test in the user's community. US-specific. Material that the relevant federal jury would find obscene under Miller may be criminal even where it depicts adult fictional characters. The standard is community-relative and prosecutorial discretion governs; the practical effect is that material at the extreme tail of fetish content (extreme violence, gore-with-sexual-context) may trigger obscenity exposure regardless of the AI provenance.

Adults using a compliant mainstream platform for fictional-character generation that respects all four red lines face no criminal exposure. The red lines are the necessary and sufficient guard rails.

How to evaluate whether a specific platform is compliant

A reader asking "is Promptchan legal" or "is Candy.ai legal" is really asking whether the platform meets the four red lines above plus the operator-side duties in their own jurisdiction. Five things to check:

• Explicit terms of service forbidding minors, real-person deepfakes, bestiality. The terms must be public, accessible, and current. Platforms with no public terms or with terms that punt on these categories should not be used.
• Documented partnership with NCMEC's CyberTipline (US) or IWF (UK). Reporting obligations under 18 USC § 2258A bind US-jurisdiction electronic communication service providers; reputable platforms surface their CyberTipline integration.
• Age verification flow for users in the eighteen US states with active statutes, the UK under OSA, and any other operator-duty jurisdiction. Geo-blocking is the alternative; either is defensible. No mechanism at all is not.
• Safety-classifier evidence. A platform that refuses obvious red-line prompts (queries for "minor", "real person ", "non-consensual") is doing the work; one that complies with such prompts is not compliant under any framework.
• Operator transparency. A real legal entity, a documented operator address, a working compliance contact email, an accessible privacy policy with data-retention specifics. Black-box operators without disclosed entities or contact channels are not platforms a careful adult should use.

The platforms we list on bestgirlfriend.ai score under our published scoring criteria, which weight Privacy & Compliance as one of the categories. That score isn't a substitute for your own due diligence on the four red lines, but it does surface operator transparency, terms-of-service depth, and verifiable corporate identity as inputs [Source: bestgirlfriend.ai: AI Companion Methodology v1.0 · verified 2026-05-17].

See Joi, image gen with Face-Sync identity-lock →

See DarLink Ai, narrative-led image gen →

What this page does not cover

This page maps the legal status of adult AI-generated content. It does not address:

• Platform-level minor-protection audits: see AI Companion Minor Protection: The bestgirlfriend.ai Red Line for our seven-criterion binary audit applied before any platform is listed.
• The federal-and-international statute map for depictions of minors specifically: see AI CSAM Laws Explained for the deep dive on § 1466A, Ashcroft, Paxton, EU CSAM Directive, UK statutes, and Germany §184b.
• State-by-state legality of AI companion apps: see Is AI Girlfriend Illegal? US State-by-State Legal Guide for the eighteen-state age-verification map.
• General AI companion safety: see Are AI Companions Safe? for the wider risk taxonomy.
• Editorial scoring methodology: see Methodology for how Privacy & Compliance is weighted across our four published scoring criteria.

Frequently asked questions

Last reviewed: 2026-05-17

Sources

1. United States, 18 USC § 1466A: Obscene visual representations of the sexual abuse of children. uscode.house.gov
2. Cornell LII: § 1466A annotation. law.cornell.edu
3. United States, 18 USC § 2258A: Mandatory reporting by ECS providers. uscode.house.gov
4. United States, PROTECT Act of 2003 (Public Law 108-21). congress.gov
5. Ashcroft v. Free Speech Coalition, 535 U.S. 234 (2002). supreme.justia.com
6. Free Speech Coalition v. Paxton, 605 U.S. ___ (2025) slip opinion. supremecourt.gov
7. Miller v. California, 413 U.S. 15 (1973). supreme.justia.com
8. Take It Down Act (Pub. L. 119-12). congress.gov
9. Texas HB 1181, 88th Texas Legislature. capitol.texas.gov
10. Utah SB 287 (no minors red line) (2023). le.utah.gov
11. United Kingdom, Online Safety Act 2023. legislation.gov.uk
12. United Kingdom, Protection of Children Act 1978. legislation.gov.uk
13. United Kingdom, Coroners and Justice Act 2009 §62 (no minors red line). legislation.gov.uk
14. European Union, Child Sexual Abuse Directive 2011/93/EU and May 2024 amendment. eur-lex.europa.eu
15. European Union, Digital Services Act, Regulation (EU) 2022/2065. eur-lex.europa.eu
16. European Union, AI Act, Regulation (EU) 2024/1689 (Article 50). eur-lex.europa.eu
17. European Commission, Guidelines on the protection of minors under Article 28 DSA (July 2025). digital-strategy.ec.europa.eu
18. Federal Republic of Germany, Strafgesetzbuch §184b. gesetze-im-internet.de
19. Canadian Criminal Code §163.1 (forbids depictions of minors). laws-lois.justice.gc.ca
20. Australian Online Safety Act 2021. legislation.gov.au
21. Australia eSafety Commissioner: Industry codes and standards. esafety.gov.au
22. US Copyright Office: AI Registration Guidance (March 2023). federalregister.gov
23. US DOJ: press release on AI-generated CSAM indictment (Wisconsin, 2024). justice.gov
24. Free Speech Coalition: Age Verification Tracker. action.freespeechcoalition.com
25. Electronic Frontier Foundation: Take It Down Act tracker. eff.org
26. NCMEC CyberTipline (public reporting). report.cybertip.org
27. Internet Watch Foundation: public report channel. iwf.org.uk
28. FTC 16 CFR Part 255: Guides Concerning the Use of Endorsements and Testimonials in Advertising. ecfr.gov

Cite this page

If you reference this page in academic, regulatory, or journalistic work, please cite as:

Joly, Alexandra (2026, May 17). Is AI-Generated Porn Legal? US, UK, EU Statute Map. bestgirlfriend.ai. https://bestgirlfriend.ai/ai-image-gen/is-ai-generated-porn-legal

Related pages

• AI CSAM Laws Explained: deep dive on § 1466A, Ashcroft, Paxton, EU and UK statutes.
• Is AI Girlfriend Illegal? US State-by-State Legal Guide: the eighteen-state age-verification map.
• AI Companion Minor Protection: The bestgirlfriend.ai Red Line: our seven-criterion platform audit.
• Are AI Companions Safe?: overall safety stance and risk taxonomy.
• Methodology: how Privacy & Compliance is weighted across our four published scoring categories.
• Editorial process: how reviews are written, fact-checked, and updated.
• Promptchan review: our scored take on the image-gen specialist.
• Best uncensored AI image generators: the full image-gen guide.

---

Last verified May 17, 2026 · See errata log for any post-publish corrections · Editor: Alexandra Joly · Methodology v1.0 · Editorial process · Affiliate disclosure