California and Europe switched AI transparency rules on.
The shared August 2 date puts content detection, machine-readable provenance, chatbot disclosure, and synthetic-media labelling into live product obligations across two major markets.
What happened
California's AI Transparency Act became operative on August 2. It covers generative AI providers with more than one million monthly users that are publicly accessible in California, requiring a free detection tool and disclosures for certain AI-generated image, video, and audio content. In the European Union, Article 50 transparency obligations also start applying on August 2: providers and deployers must disclose certain AI interactions and mark or label specified synthetic content. California's later platform and capture-device duties remain phased for 2027 and 2028, while the EU's amended rules give some older generative systems until December 2026 for marking obligations.
Why it matters
The effective date turns provenance from an optional trust feature into part of product architecture. Large providers now have to decide what metadata survives export, how users inspect it, when a chatbot reveals itself, and how a label follows content beyond the interface where it was created.
What to watch
Detection-tool accuracy, whether machine-readable marks survive screenshots and recompression, first enforcement actions, California penalty cases, the EU's December grace-period deadline, and whether providers publish usable compliance receipts rather than generic policy pages.
The caveat
The two regimes overlap in date, not in every duty. California's law focuses on covered generative-content providers and phases some platform obligations later; the EU's Article 50 rules cover a different set of provider and deployer disclosures. Formal compliance does not prove that a detector is accurate or that provenance survives real distribution.