Guide · September 2026

Instagram DM automation rules: what Meta allows, what it restricts, and how to stay inside the lines

Every week someone asks whether DM automation is "allowed". It is, and Meta publishes an API precisely so businesses can do it. What is not allowed is a short, specific list, and almost every restricted account broke one of those. Here is the rule set as it stands in 2026, then a checklist you can run any tool against.

The one rule that decides the rest: official API or not

Meta offers the Instagram Messaging API to approved apps so that businesses and creators can read and reply to DMs with software. A tool built on it connects through the "Continue with Instagram" login screen, receives a token scoped to messaging, and shows up in your Instagram settings where you can disconnect it at any time. It never sees your password.

Everything else, including browser extensions, desktop apps that ask for your login, "session cloning" and phone farms, works by pretending to be you inside the app. Meta's terms prohibit accessing Instagram through automated means outside its API, and its systems are built to detect exactly that pattern. This is the origin of most "action blocked" and "we suspect automated behaviour" notices. If a tool asks for your Instagram password, the rest of this page does not matter: it is already outside the rules.

The 24-hour reply window

Once a person sends you a message, your account (or a tool acting for it) can send replies freely for 24 hours after their most recent message. Each new message from them restarts the clock. In practice this covers almost every sales conversation, because a lead who is engaged keeps replying and keeps the window open.

The window exists so that automation is a reply, not an interruption. A lead who asked "how much?" at 11pm and got an answer at 11:01pm is the case the rule was written for; a stranger who never messaged you is the case it was written against.

The seven-day human-agent window

Meta also allows a message outside the 24-hour window when a human agent is responding, for up to seven days after the person's last message. The intended use is the follow-up a person would naturally send: "circling back on your question about Tuesday". It is sent with a specific tag and is meant for human-composed replies, not campaigns. Tools that support it (Setterr's manual follow-ups use this lane) still cannot use it to broadcast, and Meta reviews the feature before granting it to an app.

Private replies to comments and story replies

When someone comments on your post, you (or a tool) can send them one private reply in DMs for that comment, within a short window of the comment. This is the mechanism behind every "comment LINK and I'll DM it to you" automation. The reply is one message; it does not open a conversation you can keep messaging into. If the person replies to that DM, the ordinary 24-hour window starts and normal conversation follows.

A story reply is already a DM from the person, so it opens the 24-hour window like any other message.

What counts as spam

Meta does not publish a number, but the pattern it enforces against is consistent:

• Unsolicited first messages to people who never messaged you, commented, or opted in. "Mass DM" tools live here, whatever they call it.
• Repetitive content: the same message to many accounts in a short period.
• Volume spikes that do not match your account's history.
• Engagement bait such as follow-for-follow messages and automated likes or follows.

Note what is absent: replying fast, replying at 3am, replying with software, and replying to a lot of people who each messaged you first. None of that is spam. A viral reel that produces 400 inbound DMs, all answered within a minute, is fine. Sending 400 outbound DMs to people who watched the reel is not.

What gets accounts restricted, in order of frequency

1. Password-based tools and extensions. The single biggest cause. Includes tools that "only read" the inbox.
2. Outbound DM campaigns to cold accounts, especially with links.
3. Automated follows, likes and comments run alongside DM automation.
4. Sharing logins across a team on many devices and IPs at once. Not automation, but it trips the same detectors.
5. Content violations inside automated messages: prohibited products, misleading claims, impersonation.

A restriction usually arrives as a temporary block on messaging or actions; repeated hits escalate to account suspension. Recovery is slow and mostly opaque, which is why prevention is the whole game.

The 2026 picture

Two changes shape what "compliant" looks like this year. First, Meta moved Instagram messaging for businesses fully onto the Instagram API with Instagram login, so tools no longer need a linked Facebook Page to connect; if a tool still requires one, it is running an older integration. Second, AI replies are now explicitly part of the platform: Meta's own business AI features and third-party AI appointment setters both compose replies with a language model, and the rules treat an AI-composed reply inside the 24-hour window exactly like a typed one. The obligation that comes with that is on you: the AI must not make claims you cannot make, and it must not pretend to be a person if someone asks.

Compliance checklist for any DM tool

Setterr passes every row by design: Instagram login only, replies only to inbound threads inside the window, manual follow-ups on the human-agent lane, draft mode so nothing sends without your approval until you switch it off, and a live inbox where you take over with one click. That is not a feature list, it is the only way to build on the API. For the broader picture of what automation can do inside these rules, read Instagram DM automation; for setting up replies specifically, Instagram auto reply.

Frequently asked