AIBRIEF
Article 50 of the EU AI Act forces professional creators to mark realistic AI images and deepfakes starting August 2. Art, satire, fiction, and private use stay exempt.
Starting Sunday, any realistic AI image or deepfake video posted by a professional creator in the EU must carry both a machine-readable mark and, for deepfakes, a visible "AI-made" label. Satire, fiction, and purely private use stay exempt.
The trigger is Article 50 of the EU AI Act, which takes effect on August 2. Providers and deployers of generative AI systems must mark outputs in a format detectable as artificially generated or manipulated, and visibly label deepfakes and AI-generated "text in the public interest." The European Commission published draft guidelines on the Article 50(4) labelling requirement on 8 May 2026.
The rule covers professional content only. Artistic, satirical, and fictional AI work, and content made for purely private use, fall outside scope. Non-compliance carries fines of up to €15 million (roughly $16 million at recent exchange rates) or 3% of global annual turnover, per Lausen's analysis of Article 50.
Google has signed the EU AI Act's code of conduct on AI transparency. Meta has signalled intent to sign, and already runs an "AI Info" label introduced in 2024. TikTok has required AI imagery labelling for years. PetaPixel reports that both Google and Meta publicly hedged their support, warning that a growing stack of disclosure rules could confuse users.
Cross-border enforcement remains unresolved. A deepfake posted from outside the EU but reaching European viewers has no automatic remedy, leaving platforms to catch what origin-country enforcement cannot.