Automatic DM
Prewritten workflow
Disclosure point
First AI response
Recommended scope
Use broadly if unsure
Disclose when the AI conversation actually begins
Article 50 of the EU AI Act requires certain AI systems that interact directly with people to inform them that they are interacting with AI, unless that is already obvious. The European Commission says the notice should appear clearly from the start of the first interaction.
In a CommentKeyword flow, the initial keyword response and automatic DM are predefined automation. If the AI agent only starts after the recipient answers your hook, the most natural place for the disclosure is the agent's first AI-generated reply.
Automation is not automatically AI
A fixed message triggered by a keyword is not a two-way AI exchange merely because it is automated. The disclosure becomes relevant when the AI agent directly begins generating the conversation.
A natural Instagram DM flow
The disclosure does not need to appear in every message. Put it in the first response produced by the AI and then continue the conversation naturally.
Instagram conversation
Example workflow
Here's the free content calendar you asked for. What has been the biggest challenge with your Instagram content?
Staying consistent.
Thanks for sharing. Just so you know, I'm the AI assistant Laura's team uses for the first part of these conversations. What usually gets in the way of staying consistent?
Make the disclosure sound like part of the conversation
Clear does not have to mean robotic. Acknowledge the answer, identify the AI and its relationship to the business, then ask the next useful question.
Laura's coaching team
“Thanks for sharing. Just so you know, I'm the AI assistant Laura's team uses for the first part of these conversations. What usually gets in the way of staying consistent?”
Daniel's creator business
“That makes sense. Quick heads-up: I'm an AI assistant helping Daniel's team with the first part of these conversations. How long has that been your biggest challenge?”
BrightSkin
“Thanks—that's helpful. I'm BrightSkin's AI agent and the first layer of support here. Which product are you currently using?”
Northside Properties
“Great, thanks for letting me know. I'm an AI assistant for Northside Properties, helping with the first part of your enquiry. Are you looking to buy or rent?”
Peak Coaching
“I hear you. I'm Peak Coaching's AI assistant for the first part of this conversation. What result would make the next three months feel successful?”
Add the disclosure to your AI agent instructions
You can add the instruction directly when configuring your agent, or ask the Setup Assistant to update the agent for you.
Option 1
Add this to the agent instructions
In the first AI-generated response after the person answers the automatic hook, acknowledge their answer naturally and clearly explain that you are an AI assistant for [Brand]. Continue the conversation immediately afterward using the existing qualification flow and objective. Never claim or imply that you are human. Do not repeat the disclosure in later responses.
Option 2
Paste this into the Setup Assistant
Update this agent so it always discloses its AI identity in its first AI-generated response. After the person answers the automatic conversation hook, acknowledge their answer naturally and clearly explain that you are an AI assistant for [Brand]. Then continue toward the agent's existing objective. Never claim or imply that you are human, and do not repeat the disclosure in later messages. Preserve the current hook, qualification flow, objective, required fields, tone and human handoff rules.
Keep the existing agent intact
The Setup Assistant prompt explicitly preserves the hook, objective, required fields, tone and handoff rules. It changes the first AI response, not the automatic DM that comes before it.
Verify the full sequence in Test Chat
Test Chat shows an automatic keyword DM variant and its hook first. Reply as the recipient so you can review the agent's first generated response.
Pre-publish checklist
- The automatic DM and hook still appear before the AI agent takes over.
- The first AI-generated response acknowledges the person’s answer.
- That same response clearly says it comes from an AI assistant or AI agent.
- The wording identifies the brand or team the AI represents.
- The conversation continues naturally after the disclosure.
- Later responses do not repeat the disclosure.
- The agent never claims or implies that it is human.
What if the same Instagram agent reaches people worldwide?
EU AI Act requirements do not automatically apply to every interaction worldwide. CommentKeyword also cannot determine the legally relevant location of every recipient or whether a particular rule applies to your use case.
If one agent may interact with both EU and non-EU recipients and you cannot separate those audiences reliably, we recommend using the disclosure broadly for all conversations handled by that agent. It is simple, transparent and avoids asking the agent to make a location-based legal decision it cannot reliably make.
Read the EU guidance
This guide focuses on one practical Instagram DM workflow. Review the official materials and obtain professional advice for your specific circumstances.
This article provides general information, not legal advice or a complete compliance guarantee. The legal role and obligations of each party depend on the specific AI system, deployment and audience.