TECH Signal 382
EU announces new rules on AI transparency
↗
Engineers must now embed visible labels or marks on AI-generated images, audio, video, text and on AI-assisted deepfakes, and disclose when users interact with AI chatbots or agents. Providers of general‑purpose AI models must maintain training‑data summaries, publish copyright policies and face penalties of up to €15 million or 3 % of global revenue, with exemptions for personal chats and evident artistic works.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
The rules require an "AI" mark for machine‑assisted deepfake content and an "AI‑Generated" label for fully AI‑produced media such as videos, music, art or news summaries.
Emotion‑recognition and biometric‑data‑processing AI systems must also be labeled, and general‑purpose AI model operators must document information, publish training‑data summaries and implement copyright policies.
Non‑compliance can trigger fines up to €15 million or three percent of global annual revenue, while the rules exempt personal chats and evidently artistic satirical or fictional works.
THE CLUSTER