Dossier · Ongoing

The AI transparency mandate

From July's announcement to the August 2 enactment: the EU regulation and California's SB 942 require AI providers to label their content.

On August 2, 2026, labeling obligations for AI-generated content took effect simultaneously in the EU and California. Beckmann.ai first announced the EU deadline in July, once the details of Article 50 of the AI Regulation and the Digital Omnibus were settled.

On the deadline itself, it emerged that California had deliberately moved its own AI Transparency Act (SB 942) to the same day – a second jurisdiction with comparable but independent rules. This dossier tracks how both frameworks are enforced in practice.

Timeline

  1. EU AI Act: AI transparency obligations apply from August 2, 2026

    EU AI Act transparency rules apply from August 2, 2026: chatbots and AI-generated content must be labeled, fines up to €15 million.

  2. EU AI Regulation: Labeling Requirement for AI Content Starts

    From August 2, EU-wide AI labeling requirements take effect – at the same time, California also requires its own detection tool from large AI providers.

  3. Anthropic equips Claude responses with invisible watermark

    Anthropic marks Claude-generated texts and image files in a machine-readable way to meet the EU transparency obligation worldwide.

  4. Anthropic: Only complete rephrasing removes Claude watermark

    An Anthropic engineer explains to TechCrunch for the first time how the wording watermark works – and where it reaches its limits.

  5. Anthropic launches free checker tool for Claude watermarks

    The new content checker reads C2PA metadata from files and shows whether Claude generated or edited them.