AIRiskAware
AI Governance Glossary
Governance Concept

What Is AI Watermarking?

AI Watermarking is embedding a machine-detectable marker in AI-generated content so it can later be identified as synthetic.

Definition

AI Watermarkingembedding a machine-detectable marker in AI-generated content so it can later be identified as synthetic.

Watermarking helps address the provenance problem — telling whether content was produced by AI. The EU AI Act's Article 50(2) requires providers of generative systems to mark their synthetic output in a machine-readable, detectable way; under the May 2026 Digital Omnibus (provisionally agreed 7 May 2026, pending formal adoption) this obligation was given a four-month grace period to 2 December 2026, making it one of the nearest-term AI Act deadlines. Open standards such as C2PA aim to carry provenance signals across platforms. Watermarks can be removed or degraded, so they are a partial control rather than a guarantee.

Source: EU AI Act, Article 50(2); C2PA Content Provenance

Plain-language explanation

Watermarking helps address the provenance problem — telling whether content was produced by AI. The EU AI Act's Article 50(2) requires providers of generative systems to mark their synthetic output in a machine-readable, detectable way; under the May 2026 Digital Omnibus (provisionally agreed 7 May 2026, pending formal adoption) this obligation was given a four-month grace period to 2 December 2026, making it one of the nearest-term AI Act deadlines. Open standards such as C2PA aim to carry provenance signals across platforms. Watermarks can be removed or degraded, so they are a partial control rather than a guarantee.

Primary source: EU AI Act, Article 50(2); C2PA Content Provenance

Related terms

Content Provenance Deepfake Generative AI AI Transparency

See where you stand on AI governance

Take the free 7-question maturity assessment and get a personalised action plan.

Free assessment — 3 minutes →