Which Article 50 transparency duties apply to your AI?

Paragraphs 1 and 2 bind the provider: the organisation that develops the system, or has it developed, and places it on the market or puts it into service under its own name. Paragraphs 3 and 4 bind the deployer, the organisation that uses the system under its own authority. Paragraph 5 binds both.

A company that uses a vendor’s assistant under the vendor’s name is its deployer. A company that has a chatbot built for it, or builds one on a general-purpose model, and runs it under its own name is its provider (Article 3(3)), and carries paragraphs 1 and 2 itself. The safest reading is to treat the duties you can influence as yours.

Article 50 has applied since 2 August 2026. Systems placed on the market before that date have until 2 December 2026 to mark their outputs in a machine-readable way under paragraph 2. The same day, two new prohibitions in Article 5 apply.

The Commission Guidelines on Article 50 (20 July 2026) explain the duties, and the Code of Practice on AI-Generated Content (10 June 2026) describes how to mark. The Code of Practice is voluntary. A provider that does not follow it must show another way the marking duty is met.

Keep your answers: copy the link to your result or email it to yourself. They are the first lines of the classification record a regulator or an auditor will ask for.

Then test the duties that apply on the live system. A pilot sends your agent tests written from each paragraph’s text, keeps the replies as evidence and leaves the decision to your reviewer. You can export the result as a versioned and hashed Article 50 attestation, so any later change to it shows.