Guides

Plain explanations of marking AI-generated content: what Article 50 of the EU AI Act asks for, how C2PA Content Credentials and invisible watermarks work, and how to add them yourself. Each guide lists its sources.

• The 2 December 2026 deadline for marking AI-generated content

  What Article 50 of the EU AI Act asks of generative AI systems by 2 December 2026, which systems the date covers, and a technical checklist for marking.

• Article 50 of the EU AI Act: marking and labelling AI-generated content

  A plain reading of Article 50's transparency duties: which fall on providers and which on deployers, and what machine-readable marking means in practice.

• How to add C2PA Content Credentials to AI-generated images in Python

  A working Python example that signs a C2PA manifest declaring an image AI-generated, reads it back, and covers what changes for production.

• C2PA and invisible watermarks: how they differ, and why AI images use both

  C2PA Content Credentials are signed metadata; invisible watermarks live in the pixels. What each can and cannot do, and why providers combine them.

• Why Content Credentials disappear after upload, and how to keep them

  C2PA metadata is often stripped by platforms, resizers and screenshots. Where it gets lost, how to test your own pipeline, and how a watermark recovers it.

• Telling people they are talking to an AI: chatbot disclosure under Article 50

  Article 50(1) of the EU AI Act asks that people know when they interact with an AI system. What the text says, when it is obvious enough, and how to show it.

Describes technical measures. Not a legal assessment.