On 2 August 2026, a significant new phase of the EU AI Act took effect. New transparency obligations now apply to providers and deployers of certain AI systems, including chatbots, AI-generated content, deepfakes, emotion-recognition systems, and biometric categorisation tools. Under the rules, users must be informed when they are interacting with AI rather than a human, while AI-generated or manipulated audio, video, and images must be clearly labelled and include machine-readable markers where required.
The European Commission also published guidance and visual labels to support compliance. Violations may result in substantial fines under the AI Act’s enforcement framework.
Alongside the new transparency obligations, the European Commission’s AI Office officially began enforcing key parts of the AI Act from 2 August. The Commission, working with national authorities, can now investigate compliance and exercise oversight of general-purpose AI models. New operational tools were launched, including an AI Act complaints mechanism, a whistleblower channel, and reporting procedures for providers building on general-purpose AI systems. This marks the transition of the AI Act from a largely preparatory phase into active regulatory enforcement across the EU.
