AIBRIEF
The EU AI Act's transparency rules take effect today, requiring labels on deepfakes, AI images and video, certain AI written text, and customer facing chatbots, with fines for breach possible but untested.
From 2 August 2026, the EU's AI Act transparency provisions are in force. AI-generated images, video, and deepfakes circulating inside the EU must carry a visible "made by AI" label. Customer-facing chatbots have to tell users they are talking to a machine. Text covering general-interest issues that was produced by an AI without human editorial oversight must be disclosed as such, under the Commission's transparency guidelines.
The rules sit inside the EU's AI Act, the bloc's risk-based rulebook for artificial intelligence, which became law two years ago and is now entering its first enforcement phase. Their scope is narrow: every reader-facing surface AI can now generate, from a fake political video to a news explainer, has to declare itself.
Fines for breach are described as "heavy", but no specific euro amount and no first enforcement action has been published.
In Ireland, the new AI Office opens the same day to coordinate enforcement through 15 existing regulators. Minister of State Niamh Smyth calls the launch "a milestone". AI law specialist Barry Scannell of William Fry calls the rules a successor to the Turing Test's question of whether a machine can pass for human.