Patterns / AI disclosure
Telling people they're talking to an AI
The EU AI Act's transparency rules in Article 50 apply from 2 August 2026: people should be told when they're interacting with an AI system, and AI-generated content should be marked. (Providers of generating systems placed on the market before that date have until 2 December 2026 for the machine-readable marking in Article 50(2), under Regulation (EU) 2026/1744.) AiDisclosure helps with the first, as a line at the start of a thread or a badge in its header, naming the agent only as your app names it.
AiContentLabel (Pro) marks content as AI-generated or AI-modified, after the two categories in the European Commission's voluntary Code of Practice on marking and labelling AI-generated content (final, 10 June 2026): fully AI-generated and partially AI-modified. It says "AI involvement not recorded" when your app has no record. The label goes with copied text and downloads.
The components
- AI disclosureFreeAn interaction notice: a first line or a header badge.
- AI content labelProAI-generated or AI-modified, from your records; kept on copy and download (Pro).
- Agent avatarFreeThe agent's glyph, named by your app.
See it working
More in the examples: AI disclosure
Install
Each component brings in what it depends on. These are free (MIT).
Good to know
- These components help you tell people; whether what you show meets Article 50, and what else applies to your app, is your call as the deployer. This isn't legal advice, and Candor doesn't claim compliance.
- Machine-readable marking of generated files (watermarks, metadata standards) is outside a UI kit. provenanceMetadata gives simple keys for your own sidecar files.