Is AI-Generated Content Legal? A Global Regulation Map
Generating AI images or video across different countries means navigating a patchwork of laws that mostly weren't written with AI in mind. This page is a summary map, not a substitute for the deep dives — we've published detailed, sourced breakdowns of Japan, Taiwan, Thailand, and Brazil's specific rules, and this roundup adds the United States and European Union to give you a broader picture of where things stand in 2026. The short version across nearly every jurisdiction here: there's rarely a single law that regulates "AI-generated content" as its own category, what actually applies is usually existing copyright, defamation, privacy, and non-consensual-imagery law, and the clearest risk everywhere is content that depicts a real, identifiable person without consent. This isn't legal advice — for anything beyond casual personal use, talk to a lawyer licensed in your jurisdiction.
Generate fictional AI contentJapan: general laws apply, real-person content carries the most risk
Japan doesn't yet have a law that treats AI-generated content as its own legal category — what you create is generally measured against the same copyright, defamation, obscenity, and portrait-right law that would apply to a photograph or video from any other source. Japan's 2025 AI Promotion Act is a promotion-and-principles framework, not a content-restriction law, and a national AI-disclosure labeling guideline has reportedly been drafted for rollout in the second half of 2026 but isn't binding yet. The clearest risk area is content depicting a real, identifiable person without consent, which Japan's Ministry of Justice was reported to be actively studying in 2026. We go deeper on Japan's specific rules, including that study group and the labeling guideline, at /ai/ai-generated-content-japan-law.
Taiwan: a dedicated deepfake-sexual-image law, plus a brand-new AI Basic Act
Taiwan is the one jurisdiction on this list with a criminal statute written specifically for fabricated sexual imagery: Criminal Code Article 319-4, added in 2023, criminalizes creating a fabricated sexual image of a real, identifiable person without consent, punishable by up to five years in prison (seven if done for profit). Taiwan's Artificial Intelligence Basic Act, promulgated January 14, 2026, is a separate framework law setting guiding principles rather than content rules, and a narrower rule already requires online ad platforms to disclose AI-generated or deepfake material in advertising. Full detail, including how Article 319-4 is scoped, is at /ai/ai-generated-content-taiwan-law.
Thailand: general cybercrime law today, a labeling-focused AI Act still in draft
Thailand doesn't yet have a dedicated AI-content law — a draft AI Act from the ETDA opened for public hearing on July 2, 2026 and would require developers of image, audio, and video generation systems to embed machine-readable AI-generated marks, but as of this writing it's still a draft, not enacted law. Today, what actually applies to a generated image is the existing Computer-Related Crime Act, whose Section 16 criminalizes altering or creating another person's picture in a way likely to cause disrepute, insult, hate, or humiliation, alongside Penal Code defamation provisions and the Personal Data Protection Act. We cover the draft AI Act and Section 16 in more depth at /ai/ai-generated-content-thailand-law.
Brazil: no single enacted AI law, but an active non-consensual-imagery statute and a stepped-up regulator
Brazil's federal AI bill, PL 2338/2023, passed the Senate back in December 2024 and is still under committee review in the Chamber of Deputies as of mid-2026 — it isn't in force. What is already active is Penal Code Article 218-C, which criminalizes distributing a non-consensual sexual or nude depiction of a real, identifiable adult, with penalties raised to 4-to-10 years plus a fine by a December 2025 amendment, and which Brazilian courts have reportedly already applied to AI-fabricated images of real people. Brazil's data-protection regulator, the ANPD, also named AI a top enforcement priority for 2026-2027. The full breakdown, including the pending AI bill and Article 218-C's scope, is at /ai/ai-generated-content-brazil-law.
United States: a fast-growing patchwork of state laws plus a new federal takedown rule
The US doesn't have a single nationwide law regulating AI-generated content as a category — instead, as of early 2026 roughly 46 states have enacted their own laws targeting AI deepfakes, mostly focused on two areas: deceptive election-related content and non-consensual intimate imagery. At the federal level, the TAKE IT DOWN Act, signed into law in May 2025 with platform compliance requirements taking effect May 19, 2026, requires platforms to remove flagged non-consensual intimate images — including AI-generated deepfakes of a real person — within 48 hours of a verified request. A companion bill, the DEFIANCE Act, would add a federal civil damages claim for victims of sexual deepfakes; it passed the Senate again in January 2026 and remains pending in the House as of this writing.
European Union: the most comprehensive labeling regime, not a content ban
The EU AI Act takes a transparency-first approach rather than banning categories of content outright. Article 50's transparency obligations become enforceable on August 2, 2026 across all 27 member states, requiring providers of AI systems that generate image, audio, or video content to mark that output as AI-generated in a machine-readable, detectable way, and requiring deployers to disclose synthetic or manipulated content that resembles a real person, place, or event at the point a user encounters it. Non-compliance carries fines of up to €15 million or 3% of global annual turnover, whichever is higher — but the rule is about disclosure and labeling, not a prohibition on generating fictional or adult content itself.
FAQ
Is AI-generated content illegal?
Not inherently, in any of the jurisdictions covered here — most countries don't yet have a law that regulates "AI-generated content" as its own category. What tends to determine legal risk is what the content depicts (particularly whether it resembles a real, identifiable person) and how it's shared, not the fact that a generative model made it. This is a general summary, not legal advice for your specific situation.
Which countries have a law written specifically for AI deepfakes?
Taiwan is the clearest example, with Criminal Code Article 319-4 criminalizing fabricated sexual images of a real person since 2023. Thailand and Brazil have draft or pending AI bills that aren't yet enacted, and Japan's 2025 AI Promotion Act is a promotion framework rather than a content-restriction law. The EU's AI Act and a growing patchwork of US state laws focus mainly on transparency, labeling, and non-consensual imagery rather than banning AI generation outright.
Does the US have a federal law on deepfakes?
Yes, but narrowly scoped — the federal TAKE IT DOWN Act requires platforms to remove flagged non-consensual intimate images, including AI-generated ones, within 48 hours of a verified request, with compliance requirements in effect since May 2026. Most of the US deepfake legal landscape still sits at the state level, where around 46 states had enacted their own laws as of early 2026.
Does the EU ban AI-generated adult content?
No — the EU AI Act's Article 50, enforceable from August 2, 2026, is a transparency and labeling regime: it requires AI-generated image, audio, and video content to be marked as such, not that it be banned. Fines for non-compliance run up to €15 million or 3% of global turnover.
Where can I read the detailed breakdown for a specific country?
We've published deep, sourced breakdowns for Japan, Taiwan, Thailand, and Brazil — see /ai/ai-generated-content-japan-law, /ai/ai-generated-content-taiwan-law, /ai/ai-generated-content-thailand-law, and /ai/ai-generated-content-brazil-law. This page is a summary map across all of them plus the US and EU; it isn't a substitute for legal advice.
Popular NSFW templates
