Watermark Management System
For Compliance with the EU Artificial Intelligence Act - Article 50
The Challenge: Transparency in the Age of Generative AI
The EU Artificial Intelligence Act (Regulation (EU) 2024/1689) introduces groundbreaking transparency obligations under Article 50. These requirements apply to providers and deployers of generative AI systems and are designed to protect individuals from being misled by AI-generated or manipulated content.
Key Regulatory Drivers
- Article 50(2) - Providers must ensure AI-generated outputs are marked in a machine-readable format and detectable as artificially generated.
- Article 50(4) - Deployers must clearly disclose deep fakes and certain AI-generated text published for public information.
- Article 50(5) - Information must be provided in a clear, distinguishable manner at the point of first interaction or exposure.
- Code of Practice (2026) - Operationalises these obligations through concrete Commitments and Measures for marking, detection, and labelling.
Traditional point-in-time audits are insufficient for AI systems that continuously generate new content. Organisations need ongoing governance, evidence collection, and the ability to demonstrate compliance to market surveillance authorities and other stakeholders.
The Solution: A Watermark Management System
The Watermark Management System is a comprehensive governance and compliance platform designed specifically to help organisations meet the transparency obligations of the EU AI Act for generative AI content. It provides the structured processes, evidence management, and audit readiness required to demonstrate compliance with Article 50 and the associated Code of Practice.
What the System Delivers
For Providers of Generative AI Systems
- Structured support for multi-layered marking approaches (metadata + watermarking)
- Governance framework for detection solution provisioning and access management
- Documentation of technical solutions meeting effectiveness, interoperability, robustness, and reliability criteria
- Ongoing monitoring of marking and detection performance
- Ready-to-use evidence packages for market surveillance authorities
For Deployers of AI-Generated Content
- Guidance and record-keeping for perceptible labelling (EU icon compliance)
- Support for human review and editorial control policies
- Internal compliance process documentation
- Traceable records of disclosure implementation
Core Value Proposition
Rather than treating transparency as a one-time technical implementation, our system establishes transparency as an ongoing, auditable business process. This transforms compliance from a burden into a demonstrable organisational capability that builds trust with regulators, customers, and the public.
Key Capabilities
- Centralised repository for all transparency-related decisions, policies, and evidence
- Clear separation of provider and deployer obligations with tailored workflows
- Version-controlled documentation aligned with the Code of Practice
- Workflow-driven approval and sign-off processes
- Framework for implementing and documenting machine-readable marking techniques
- Support for digitally signed metadata and imperceptible watermarking approaches
- Detection result recording and performance monitoring
- Quality criteria tracking (effective, reliable, robust, interoperable)
- End-to-end process from AI system registration through ongoing monitoring
- Integration points with risk management, change control, and post-market surveillance
- Clear audit trail from obligations to evidence
- One-click generation of structured evidence packages for market surveillance authorities
- Support for both minimum and full QMS-integrated deployment profiles
- Traceable, signed records suitable for regulatory scrutiny
- Ongoing performance tracking of transparency measures
- Gap analysis against evolving Code of Practice requirements
- Corrective and preventive action workflows
Alignment with the EU AI Act & Code of Practice
The Watermark Management System is purpose-built to support the Commitments and Measures defined in the Code of Practice on Transparency of AI-Generated Content (2026).
| CoP Area | System Support |
|---|---|
| Section 1 - Providers | Multi-layer marking governance, detection solution documentation, quality criteria tracking, interoperability planning, MSA cooperation records |
| Section 2 - Deployers | EU icon implementation records, human review policies, internal compliance processes, disclosure traceability |
| Measure 4.1 - Compliance Process | Documented, version-controlled compliance processes with real examples and audit trails |
| Measure 4.2 - Testing & Verification | Red-teaming records, benchmark results, performance monitoring, corrective action tracking |
| Measure 4.4 - MSA Cooperation | Pre-prepared evidence packages, documented processes, and rapid response capability |
The system is designed to evolve alongside the Code of Practice and related Guidelines, ensuring long-term compliance sustainability.
How It Works - High-Level Workflow
The Watermark Management System follows a phased, auditable workflow that transforms regulatory obligations into operational reality.
Phase 1: Registration & Scoping
- Register the AI system and capture transparency-relevant characteristics
- Determine whether the system falls under provider, deployer, or both obligations
- Define the scope of marking, detection, and labelling requirements
Phase 2: Governance Setup
- Establish the structured compliance record aligned with Code of Practice Commitments
- Document technical solutions, policies, and internal processes
- Set up monitoring and evidence collection mechanisms
Phase 3: Operationalisation & Monitoring
- Implement and document marking and detection capabilities
- Conduct ongoing performance monitoring and red-teaming
- Maintain traceability between risks, measures, and evidence
Phase 4: Evidence & Authority Engagement
- Generate structured, signed evidence packages on demand
- Respond efficiently to market surveillance authority requests
- Demonstrate continuous compliance and improvement
The system supports both "minimum" deployment (focused on transparency demonstrability) and "integrated" deployment (full QMS embedding with change control, nonconformity management, and pervasive governance controls).
Benefits & Value
For Compliance & Legal Teams
- Reduced risk of regulatory non-compliance and associated penalties
- Clear, auditable evidence ready for market surveillance authorities
- Streamlined response to regulatory inquiries
For Product & Engineering Teams
- Structured framework to document and govern transparency implementations
- Clear visibility of obligations across the AI system lifecycle
For Executive Leadership
- Demonstrable organisational capability in AI transparency
- Reduced reputational and financial risk
- Foundation for trustworthy AI practices
Deployment Flexibility
- Minimum Profile: Rapid deployment focused on transparency demonstrability with lighter process overhead.
- Integrated Profile: Full embedding into enterprise QMS with change control, nonconformity management, and pervasive governance.
Ready to strengthen your AI transparency compliance?
Contact us to discuss how the Watermark Management System can be tailored to your organisation's generative AI portfolio and compliance roadmap.
