Topic dossier
When people have to be told they are dealing with AI, and content has to be marked as model-generated — chatbot disclosure, content labelling and training-data transparency. 9 obligations across 7 jurisdictions — 7 in force. Next dated deadline: 2 Aug 2026.
Transparency is the most common thread running through AI law. Three duties recur: telling a person when they are interacting with an AI system, marking content that an AI generated, and — increasingly — disclosing what data a model was trained on. The instruments below each carry one or more of these duties, tracked to their primary sources.
Binds Providers & deployers of interactive, synthetic-content or biometric AI. Disclosure of AI interaction; marking of AI-generated content.
Stated maximum penalty — Up to 3% turnover or €15M
Binds Operators of companion-chatbot platforms available in California. AI-status disclosure + self-harm protocols.
Stated maximum penalty — Private right of action
Binds Developers of generative AI systems made available to Californians. Public dataset-summary disclosure for generative AI offered to Californians.
Stated maximum penalty — Civil enforcement
Binds Covered GenAI providers with >1M monthly users accessible in California. AI-detection tool + content provenance for >1M-user providers.
Stated maximum penalty — Civil penalties per violation/day
Binds AI-content service & propagation platforms, app stores, and users. Explicit (visible) and implicit (metadata/watermark) labels on AI-generated content.
Stated maximum penalty — CAC administrative penalties
Binds AI business operators offering AI products/services in Korea (extraterritorial). Pre-notify users that a service uses AI; label generative and realistic synthetic outputs.
MSIT enforcement grace period through ~2026.
Stated maximum penalty — Admin fine up to ₩30M
Binds Providers / deployers of generative AI and user-facing AI systems. Machine-readable labels on AI media; disclose when users interact with AI; deceptive deepfakes banned.
Stated maximum penalty — Admin fines (decree-set)
Binds Intermediaries, significant social-media intermediaries (5M+ users), GenAI tool providers. Permanent labels on AI-generated media + rapid takedown; significant-platform traceability.
Stated maximum penalty — Loss of safe harbour; IT Act offences
Binds Organisations making automated decisions using personal information in Quebec. Right to be informed + disclosure of key factors for automated decisions using personal info (Quebec).
Stated maximum penalty — AMPs up to C$10M / 2% turnover
Several laws now require it. The EU AI Act’s Article 50, California’s SB 243 companion-chatbot rules, South Korea’s AI Basic Act and Vietnam’s Law on AI all require users to be told that they are dealing with an AI system rather than a human.
A duty to publish a summary of the data used to train a model. California’s AB 2013 requires generative-AI developers to post a dataset summary; the EU AI Act requires GPAI providers to publish a sufficiently detailed summary of training content.
Quebec’s Law 25 gives individuals a right to be informed of, and to understand the main factors behind, an automated decision made about them using personal information; Colorado’s AI Act adds consumer notice and appeal rights for consequential automated decisions.
We currently track transparency & disclosure obligations across 7 jurisdictions: European Union, United States, China, South Korea, Vietnam, India and Canada. Each is dated and linked to its primary source on this page.
Not legal advice. Each obligation links to its primary source and carries the date it was last checked; verify the legal text before relying on it.