Record STMT-W9322F2Advisory

Generative AI and watermarking

Report State lawEuropean Unionunknown

This briefing by the European Parliamentary Research Service examines the role of generative AI and watermarking techniques in establishing transparency and content provenance. It reviews current technological limitations of watermarking, international approaches in China, the US, the G7, and the European Union, and discusses open policy questions around copyright and standardisation.

Summary and content fields are drawn from the document text and are pending editorial review.

Who issued it

European ParliamentLegislature

Lead issuer

Legal forceAdvisory

Under whose authorityState law

Judged fromfull text

Who it governs

LegislatorsRegulatorsAI developers and providersResearchers and academics

European Parliament members, staff, policymakers, and researchers seeking an overview of generative AI watermarking and regulatory frameworks.

Sectors

Telecommunications and digital services · generative artificial intelligenceResearch and science

What the document contains

What's covered

The document covers the technical definition and mechanisms of generative AI watermarking, its benefits for content authentication and copyright protection, its current technical limitations, and various regional regulatory approaches including the EU AI Act, Chinese guidelines, US executive orders, and G7 principles.

What's missing

It does not contain binding legislative text, formal enforcement penalties, or specific technical specifications for proprietary watermarking algorithms.

Key takeaways

Generative AI creates challenges in distinguishing human-generated content from synthetic outputs, driving the need for reliable watermarking and provenance mechanisms.
Current state-of-the-art watermarking techniques face significant technical limitations, including vulnerability to backdoor attacks, false positives, and lack of standardisation.
Major jurisdictions including China, the US, the G7, and the European Union are actively introducing regulatory and policy requirements for labelling and watermarking AI content.
Watermarking can help address copyright compliance, data monitoring, and the prevention of AI-generated misinformation and deepfakes.
Establishing a robust watermarking environment requires complementary measures such as mandatory documentation, foundation model pre-release testing, and interdisciplinary research.

Outline

· Generative AI: need for transparency
· AI watermarking techniques
· AI watermarking benefits
· AI watermarking limitations and drawbacks
· AI regulation and watermarking
· AI watermarking implementation and open questions

IssuedDecember 2023

Length3,456 words

FormatNot recorded

Scoperegional

How it scores

Pending review

Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.

Accountability

2/5

Clear responsibility for AI outcomes

Self-disclosing AI machines could be the solution to ensure AI's transparency and accountability.

Setting up a robust watermarking environment

screening score · 2026-10-23

Human Dignity

1/5

Respect for the inherent worth of persons

S. Gregory and R. Vazquez LLorente, Regulating Transparency in Audiovisual Generative AI: How Legislators Can Center Human Rights

Setting up a robust watermarking environment

screening score · 2026-10-23

Human Oversight

0/5

Humans remain in control of consequential decisions

screening score · 2026-10-23

Transparency

5/5

Openness about how AI systems work and are used

Generative AI: need for transparency

IN THIS BRIEFING

screening score · 2026-10-23

Enforcement

1/5

Consequences and compliance mechanisms with real force

There are also calls on Congress to pass legislation on AI-generated content in 2024 to ensure the watermarking mechanism rules are enforced.

AI regulation and watermarking

screening score · 2026-10-23

Topics

generative_ai watermarking transparency copyright standardisation