Deepfakes: the line so far

Every generative AI service and content platform operating in China now has to apply both visible labels and machine-readable provenance, a model later echoed in India and the EU.

It shows a central bank using supervised sandboxes, rather than new rules, to shape how banks deploy GenAI and counter deepfake fraud.

It is one of the first comprehensive AI laws with binding duties to take effect outside the EU, and a test of how such a law is phased in for global providers.

Platforms and AI tools serving Indian users must build labelling and provenance into products or risk losing safe-harbour protection.

It gives an evidence base, beyond rhetoric, for what China's military is actually buying, which informs US export controls and allied threat assessments.

Platforms and generative AI providers operating in Brazil face binding election-specific labelling, blackout and removal duties during the October 2026 general election.

Hollywood's main unions have locked in AI consent and use limits until 2030, as workers in other sectors push for their own AI protections.

Signing is the simplest way for model providers and publishers to evidence compliance with EU labelling duties now in force.

Providers and public bodies deploying high-risk AI in the EU gain 16-24 months, while the new prohibitions and watermarking deadline arrive in December 2026.

It is one of the first concrete large-scale answers to the EU's AI-content marking duties, and shows the limits: short or heavily edited text is hard to detect, and the detector is not public.