Which countries have binding AI laws?
Three jurisdictions now enforce rules written specifically for AI. The European Union's AI Act, Regulation (EU) 2024/1689, sets risk-based duties for AI systems and general-purpose models across all 27 member states. China regulates generative AI services and AI-generated content through administrative measures issued by the Cyberspace Administration of China. South Korea's AI Basic Act took effect on January 22, 2026, adding labelling and high-impact AI duties with administrative fines.
A larger group has chosen not to pass a horizontal AI statute. The UK, Canada, Australia and India apply privacy, consumer protection, online safety, copyright and sector rules to AI, sometimes with new AI-specific amendments to those older laws. India's February 2026 amendment to its IT intermediary rules is the clearest example: it is binding and aimed squarely at AI-generated media, but it sits inside a platform law rather than a standalone AI act.
Japan sits in between. Its 2025 AI Promotion Act is a statute, but it carries no monetary penalties and works mainly by directing government strategy and asking businesses to cooperate. The practical test for any reader is simple: ask whether a regulator can fine you for how you build or deploy an AI system. In late 2026, the answer is clearly yes in the EU, China and South Korea, and mostly no elsewhere outside sector rules.
AI regulations around the world: the table
The table compares the main AI law or framework in each jurisdiction we track, with its status as of November 9, 2026. "Binding" means a regulator can impose penalties under the instrument itself.
| Jurisdiction | Law or framework | Binding | Status | In force from | Regulator |
|---|---|---|---|---|---|
| European Union | AI Act, Regulation (EU) 2024/1689, as amended by Regulation (EU) 2026/1744 | Yes | In force, phased application | August 1, 2024 (bans from February 2, 2025; most rules from August 2, 2026; high-risk from December 2, 2027 and August 2, 2028) | AI Office and national authorities |
| China | Interim Measures for Generative AI Services; AI content labelling measures; amended Cybersecurity Law | Yes | In force | August 15, 2023; September 1, 2025; January 1, 2026 | Cyberspace Administration of China and co-regulators |
| South Korea | AI Basic Act and Enforcement Decree | Yes | In force; fines deferred for at least one year | January 22, 2026 | Ministry of Science and ICT |
| Japan | AI Promotion Act | No penalties | In force | June 4, 2025 (main provisions); September 2025 (strategy chapters) | AI Strategy Headquarters (Cabinet) |
| India | IT Rules amendment on synthetically generated information | Yes, for platforms | In force | February 20, 2026 | Ministry of Electronics and IT |
| Canada | No AI statute; AIDA died in 2025; privacy bill C-36 tabled June 2026 | No | Bills pending | Not applicable | Existing privacy and competition regulators |
| Brazil | PL 2338/2023 AI framework bill | Not yet | Passed Senate; in Chamber committee | Not applicable | Not yet set by law |
| United Kingdom | No AI statute; sector regulators apply existing law | No | No AI bill in the May 2026 King's Speech | Not applicable | Sector regulators |
| Australia | National AI Plan (December 2025); mandatory guardrails dropped | No | Policy only | Not applicable | Existing regulators; advisory AI Safety Institute |
| Council of Europe | Framework Convention on AI (CETS 225) | Yes, for parties | Open for signature; not in force | Not yet | Parties' national authorities |
Sources for each row are linked in the country sections below. For the EU's own timeline and the Article 50 transparency rules, see our EU AI Act content rules page; for US state law, start at our legislation index.
European Union: the AI Act
The AI Act entered into force on August 1, 2024 and applies across all 27 member states. It sorts AI systems into four risk levels, bans a list of practices outright, imposes conformity duties on high-risk systems, adds disclosure duties for chatbots and generated content, and sets separate rules for general-purpose AI models. The European Commission's AI Act page lists the application dates: bans and AI literacy from February 2, 2025, general-purpose model rules from August 2, 2025, and general application from August 2, 2026.
The Digital Omnibus on AI changed that schedule. Regulation (EU) 2026/1744 was published on July 24, 2026 and entered into force on July 27, 2026. It moves stand-alone (Annex III) high-risk obligations to December 2, 2027 and product-embedded (Annex I) obligations to August 2, 2028, gives generative systems already on the market until December 2, 2026 to meet the watermarking duty, and adds a ban on AI systems that generate non-consensual intimate imagery or child sexual abuse material. Fines for the most serious breaches reach EUR 35 million or 7 percent of worldwide annual turnover, whichever is higher, under Article 99 of the official text.
Asia: China, South Korea, Japan and India
China
China regulates AI through targeted administrative measures rather than one code. The Interim Measures for the Management of Generative AI Services took effect on August 15, 2023 and apply to generative text, image, audio and video services offered to the public in mainland China. The Measures for Labelling AI-Generated and Synthetic Content, issued in March 2025 with a mandatory national standard, took effect on September 1, 2025 and require both visible labels and embedded metadata labels. China also amended its Cybersecurity Law in October 2025, adding its first express AI provision with effect from January 1, 2026.
South Korea
South Korea's Framework Act on the Development of Artificial Intelligence and the Establishment of a Foundation for Trust, usually called the AI Basic Act, took effect on January 22, 2026 together with its Enforcement Decree. According to Cooley's January 2026 summary, it requires notice and labelling for generative AI output, imposes explainability, human oversight and documentation duties on high-impact AI, designates models trained above 1026 floating-point operations as high-performance AI, and sets administrative fines of up to KRW 30 million. Foreign operators above revenue or user thresholds must appoint a local representative. The US International Trade Administration reports that the ministry will generally defer fact-finding investigations and fines for a grace period of at least one year.
Japan
Japan's Act on the Promotion of Research, Development and Utilization of AI-Related Technologies was passed by the Diet in May 2025, and most of its provisions took effect on June 4, 2025. White & Case notes that it imposes no monetary penalties; instead it creates an AI Strategy Headquarters in the Cabinet and a duty on businesses to cooperate with government measures. The Cabinet approved Japan's first AI Basic Plan in December 2025. Japan therefore has an AI law, but not an AI rulebook with sanctions.
India
India has no standalone AI statute. Its binding AI rules arrive through the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules. An amendment notified on February 10, 2026 and in force from February 20, 2026 defines synthetically generated information, requires platforms to label it prominently, and shortens takedown timelines for unlawful content. The duties fall on intermediaries rather than on AI developers as such.
The Americas: Canada and Brazil
Canada
Canada's proposed Artificial Intelligence and Data Act, part of Bill C-27, died when Parliament was prorogued. The Logic reports that AI Minister Evan Solomon has said he will not reintroduce it. The government's June 2026 approach relies on privacy reform and online safety legislation instead: Bill C-36, tabled in June 2026, would update federal privacy law and give people a way to ask companies to delete deepfakes using their likeness. Michael Geist's analysis of the June 2026 AI strategy describes it as relying on those bills rather than AI-specific regulation. As of November 9, 2026, Canada has no federal AI statute in force.
Brazil
Brazil's PL 2338/2023 is a full AI framework bill modelled on risk tiers. It passed the Federal Senate in December 2024 and was sent to the Chamber of Deputies, where a special committee has been considering it. The Rio Times, citing the Chamber's processing record checked on November 5, 2026, reports that the committee is still waiting for the rapporteur's opinion and no vote is scheduled. Brazil's 2026 general election calendar makes a final vote before the end of the year uncertain.
UK and Australia: regulating AI through existing law
The United Kingdom has repeatedly signalled AI legislation and then deferred it. The May 2026 King's Speech did not include a standalone AI bill; the government instead announced a Regulating for Growth Bill built around an AI Growth Lab sandbox. In practice, UK AI rules come from data protection law enforced by the Information Commissioner's Office, competition and consumer law, the Online Safety Act and sector regulators.
Australia moved the same way in December 2025. Its National AI Plan dropped the mandatory guardrails for high-risk AI that the government had proposed in 2024 and relies on existing laws, updated case by case. IAPP's summary notes the plan funds an AI Safety Institute to monitor risks and advise regulators. Australia's binding AI rules therefore remain the privacy, consumer, online safety and anti-discrimination laws that already applied.
Is there a global AI treaty? The Council of Europe convention
The Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (CETS 225) is the first legally binding international AI treaty. It was opened for signature in September 2024 and is open to Council of Europe members, non-member states that helped draft it, and the European Union. It requires parties to keep AI activity consistent with human rights, democracy and the rule of law, with risk assessment, transparency and remedy obligations applied through national law.
Signature is not ratification. North Macedonia's signature in May 2026 brought the number of signatories to 20, and the European Union ratified the convention on May 15, 2026. Under its final clauses, published in the treaty text, the convention enters into force only after five ratifications, including at least three by Council of Europe member states. That threshold had not been met in any source we found as of November 9, 2026, so the treaty binds no one yet.
Companies operating across borders face binding AI-specific rules in only three places as of November 9, 2026: the European Union, China and South Korea. Japan has an AI statute with no penalties, while the UK, Canada, India and Australia rely on existing law and targeted rules. Brazil's AI bill is still in its lower house, and the Council of Europe AI treaty is not yet in force.
What changed
- 2026-11-05: Brazil's PL 2338/2023 still awaiting the rapporteur's opinion in the Chamber of Deputies committee, with no vote scheduled.
- 2026-07-27: EU Digital Omnibus on AI (Regulation (EU) 2026/1744) entered into force, delaying high-risk obligations to December 2027 and August 2028.
- 2026-06-15: Canada tabled Bill C-36 on privacy, including deletion rights covering deepfakes; no AI statute proposed.
- 2026-05-15: European Union ratified the Council of Europe Framework Convention on AI.
- 2026-02-20: India's IT Rules amendment on synthetically generated information took effect.
- 2026-01-22: South Korea's AI Basic Act and Enforcement Decree took effect.
- 2026-01-01: China's amended Cybersecurity Law, with its first AI provision, took effect.
- 2025-12-02: Australia released its National AI Plan, dropping proposed mandatory guardrails.
Frequently asked questions
Which countries have AI laws?
As of November 9, 2026, the European Union, China and South Korea enforce binding AI-specific rules. Japan has an AI statute without penalties. India regulates AI-generated content through platform rules. The UK, Canada and Australia rely on existing privacy, consumer and online safety law, and Brazil's AI bill is still pending.
Is there a global AI treaty?
The Council of Europe Framework Convention on AI (CETS 225) is the first binding international AI treaty. It had 20 signatories by May 2026 and the EU ratified it on May 15, 2026, but it needs five ratifications, including three Council of Europe members, before it enters into force.
Which country has the strictest AI regulation?
The EU AI Act is the broadest binding regime, with outright bans, conformity duties for high-risk systems and fines up to 7 percent of worldwide turnover. China's rules are narrower in scope but tightly enforced through content and labelling controls on generative AI services.
How do global AI regulations compare with the US?
The US has no economy-wide federal AI statute. Federal rules are targeted, such as the TAKE IT DOWN Act on intimate images, and most AI-specific law comes from states. That makes the US closer to the UK and Australia than to the EU's single horizontal regime.
Sources
- EUR-Lex: Regulation (EU) 2024/1689 (AI Act)
- European Commission: AI Act policy page
- Council of Europe: EU ratifies the Framework Convention on AI
- Council of Europe: CETS 225 treaty text
- White & Case: EU AI Omnibus enters into force
- Cooley: South Korea's AI Basic Act overview
- Herbert Smith Freehills Kramer: China's AI content labelling rules
- White & Case: Japan's first AI legislation
- Freshfields: India's 2026 IT Rules amendment
- The Rio Times: Brazil's AI bill status
- IAPP: King's Speech 2026 and AI
- MinterEllison: Australia's National AI Plan
This page is journalism and commentary, not legal advice. AI law changes quickly; check the primary sources above and consult qualified counsel for your situation.