Article 50 Disclosure Examples for AI Chatbots and AI Content

The obligation, in plain terms

Article 50 requires that natural persons are informed when they interact with an AI system, unless it is obvious from the circumstances. It also requires labelling of certain AI-generated or manipulated content.

Effective date: Art. 50 transparency applies from 2 August 2026 (unchanged by the Digital Omnibus). That is earlier than standalone Annex III high-risk duties (2 December 2027).

This is a drafting guide — not a finding that your product is in or out of scope.

Chatbots and support AI

Minimum viable disclosure, shown at the start of the conversation:

"You're chatting with an AI assistant. Responses may be inaccurate. You can request a human agent at any time."

What usually fails: burying the notice in a help centre or Terms of Service. Disclosure needs to be at the point of interaction — in the chat UI itself.

More variants: AI disclosure notice template · Article 50 checklist

AI-generated content (marketing copy, reports, summaries)

Minimum viable disclosure, shown adjacent to the content:

"This summary was generated by AI based on [source]. Reviewed by [role] before publishing."

or

"AI-generated — not reviewed by a human."

Common mistake: labelling the feature ("AI Summarizer") without labelling each output that can be copied, exported, or shared.

Recommendation / scoring engines

Near the recommendation:

"Recommended by AI based on your usage patterns. [Why am I seeing this?]"

If the output affects a person materially (candidate ranking, credit pre-check, pricing), pair disclosure with a path to explanation or human review. Art. 50 is the notice; oversight and GDPR Art. 22 are separate questions.

AI agents taking autonomous actions

In the action's output (email sent, ticket updated):

"This message was sent by an AI agent on behalf of [company]. Reply to reach a human."

Principle: the person on the receiving end should be able to tell without digging.

What "obvious from context" usually does not cover

Spam filters, internal analytics, and back-office automation with no human-facing interaction are often argued as out of scope for interaction disclosure. The moment a person reads, receives, or acts on AI output directly, that argument gets harder.

10-minute fix

1. Add one clear line to your primary AI touchpoint.
2. Match the wording to real product behaviour (including human escalation if you promise it).
3. Get a free readiness brief for your product to list remaining Art. 50 gaps with sources.

Also see: AI chatbot checklist · Transparency obligations overview · Article 50 cluster

Not legal advice. Confirm whether Art. 50 applies to each interaction type with counsel.