Shopify Inbox mixes two modes, and the Article 50(1) line runs exactly between them.
AI-suggested replies that you review and send yourself are human agent-assist: the AI drafts, a human decides and sends. The Commission's draft guidelines put this configuration outside the chatbot disclosure duty entirely. Article 50(1) is triggered by fully automated responses, not by a human using AI as a drafting tool. If every message to a customer passes through a person who actually reads it before sending, you don't need an AI notice for that flow.
Fully automated answers are the covered case: instant answers, or an AI assistant replying to shoppers on its own under your store's name. Your store, not Shopify, is the brand the shopper sees, which generally makes you the provider of that chatbot under the Act, with the duty to tell shoppers they're chatting with AI at or before the first automated message.
The honest complication is that many stores run both modes in the same chat window, and the customer can't tell which mode produced a given reply. The clean solution is also the cheapest one: put a short AI notice on the chat surface anyway (first message and/or a persistent badge). It fully covers the automated flow, costs you nothing on the assisted flow, and spares you keeping a legal analysis in sync with a settings toggle.
Keep a dated screenshot of the disclosure and a note of which Inbox automation features you have enabled. That pair is exactly the evidence an authority or a large B2B customer would ask for first.
Not sure how this applies to your specific setup? The free 2-minute check tells you which of the four Article 50 duty areas likely apply to your company. No email required.
Take the free check Generate disclosure text