AI Act Article 50(1) chatbot notice for an agent built on Sume
Article 50(1) asks providers to make sure people know they are talking to AI. If you front Sume Agent Completions with a chat box, the notice is yours to add.
Article 50(1) says providers of AI systems meant to interact directly with people must design them so those people are informed they are interacting with an AI system, unless that is obvious from the context (Article 50, read 2026-10-02). If you put a chat box in front of a Sume Agent Completion, the notice is something you build into that chat box. Sume's API does not add one for you.
What paragraphs 1 and 5 require
The obligation applies from 2 August 2026, the date the Commission gives for Article 50 becoming applicable (Commission page, read 2026-10-02). Two paragraphs matter for a chat product.
| Paragraph | What it says |
|---|---|
| 50(1) | Persons are informed they are interacting with an AI system, unless obvious from the context |
| 50(5) | Information given in a clear and distinguishable manner, at the latest at first interaction or exposure |
| 50(5) | The information must conform to applicable accessibility requirements |
How a Sume agent fits
Sume's Agent Completions docs say a completion is not a synchronous chat completion: the create call returns 202 with a receipt that you poll, and streaming and a sync OpenAI-compatible wire are not available yet. The request borrows the messages[] shape, but the response is a run receipt.
That means any chat window is your own front end calling the API from your backend. You decide what the user sees on first load, so the AI notice is part of your design, not an API field. The same applies when you run an agent from a scheduled action: nobody is watching, so the notice must live wherever the person eventually reads the result.
A first-load notice that works
- Show the notice before the first message, not after the first reply, to meet the first-interaction timing in 50(5).
- Use plain words such as "You are chatting with an AI assistant" in text that screen readers can read.
- Keep it visible when the agent hands off a task; a receipt that arrives later should still be marked as automated.
- Do not rely on the obvious-from-context exception for a branded persona with a human name and photo.
What this page does not decide
Whether you are the provider, the deployer, or both depends on your product and is a legal question. Sume's docs do not claim to make a product Article 50 compliant. They describe an API that returns a receipt and results; the transparency design stays with you.
Sources
Related posts
More in Agents
- GitHub Actions schedule delayed at the top of the hour: what to do
GitHub says scheduled workflows can be delayed, notably at the start of every hour. Offset the cron minute and call a Sume Scheduled run with a dated key.
- GitHub Actions schedule disabled after 60 days: a Sume Scheduled fix
A public repo's scheduled workflows are disabled after 60 days without activity. How Sume Scheduled's active and inactive status differs.
- Higgsfield MCP has no credit cap: cap spend with Sume max_spend_usd
Higgsfield's MCP guide says there is no built-in credit spending cap. Sume's MCP tools take dry_run and max_spend_usd. How they work, and what they skip.
- hypit Understand order: one probe, then parallel batches
Sume's hypit Understand order: probe alone, then transcribe, boundaries and tiles in one batch, then notes. Which verbs wait on the transcript.
Written by Sume