Watermarks and Metadata: How to Actually Comply With the EU AI Act's Article 50 Transparency Rules

2026-06-15 · Source: Global Privacy Laws & Compliance Frameworks | ComplianceHub.Wiki · Field: Legal & Regulatory — Compliance & Risk Management, Regulatory Affairs & Government Relations, Artificial Intelligence & Machine Learning · Depth: Intermediate, long

Summary

The European Commission and European AI Office published the final Code of Practice on Transparency of AI-Generated Content on June 10, 2026, providing concrete guidance for complying with the EU AI Act's Article 50 transparency rules. Article 50, which becomes legally binding on August 2, 2026, mandates that providers of generative AI systems mark outputs in a "machine-readable format" and deployers label deepfakes "clearly." The Code, while voluntary, outlines two core marking mechanisms for providers: digitally signed metadata, often using C2PA standards, and imperceptible watermarking for durability. It also details deployer duties for visible labelling of deepfakes and public-interest text, including using standardized EU icons. A transitional period extends to December 2, 2026, for systems placed on the market before August 2, 2026. Non-compliance can result in administrative fines up to 15 million euros or 3% of total worldwide annual turnover.

Key takeaway

For Directors of AI/ML and AI Engineers deploying generative AI in the EU, you must immediately assess your systems against the EU AI Act's Article 50 and the new Code of Practice. Your teams should implement both signed metadata and imperceptible watermarking for provider systems, and design visible labels using EU icons for deployer obligations. Ensure your terms of service prohibit marking removal and track the December 2, 2026, transition for existing systems to avoid significant fines.

Key insights

The EU AI Act's Article 50 transparency rules are operationalized by a new Code of Practice endorsing specific marking and labelling mechanisms.

Principles

• Adherence to the Code of Practice signals good-faith compliance.
• Provider marking and deployer labelling are sequential duties.
• Metadata and imperceptible watermarking are complementary.

Method

Providers should implement digitally signed metadata (e.g., C2PA) and imperceptible watermarking, making detection tools freely available. Deployers must surface these markings visibly using EU icons.

In practice

• Inventory all generative AI systems in EU operations.
• Design visible labels for deepfakes using standardized EU icons.
• Update terms of service to prohibit marking removal.

Topics

• EU AI Act
• AI Transparency
• Content Provenance
• Digital Watermarking
• C2PA Standard
• Deepfake Labelling

Best for: Legal Professional, Director of AI/ML, AI Engineer

Related on AIssential

• EU Commission Publishes AI Transparency Code of Practice — The EU's AI Transparency Code provides a voluntary framework for marking and disclosing AI-generated content to ensure compliance with the AI Act.
• Supporting Europe’s work in ensuring a trustworthy AI ecosystem — Effective AI content provenance requires multi-layered technical approaches and broad ecosystem collaboration to ensure transparency.
• EU AI Act Article 50: Transparency Rules for Businesses — The EU AI Act's Article 50 mandates transparency for diverse AI systems, requiring users to know when they interact with or consume AI-generated content.
• Advancing Transparency of AI-Generated Media in the EU Code of Practice — Effective synthetic media transparency requires a multi-layered approach combining technical markings with clear user-facing disclosures.
• What the EU’s new Code of Practice means for AI-generated content transparency — The EU's Code of Practice establishes a "hallmarking" system for AI-generated content to ensure transparency and trust.
• AI Transparency Deadline Approaching — The EU AI Act mandates strict transparency and disclosure for AI systems, particularly concerning human interaction, generated content, and biometric data.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Global Privacy Laws & Compliance Frameworks | ComplianceHub.Wiki.