ChatGPT Joins Slack and Microsoft Teams: Mention It in Any Thread

OpenAI put ChatGPT inside the two places enterprise conversation actually happens: at DevDay 2026 it launched @ChatGPT mentions for Slack and Microsoft Teams, letting anyone in a channel, thread, or direct message request work by tagging the assistant. ChatGPT can use the connected tools it has been authorized for, teammates can pile context into the same thread, and — the commercially notable detail — colleagues can participate without each needing an individual subscription.

This piece covers how the mention model works, why thread-level participation differs from the chatbot-in-a-sidebar pattern, what the no-seat participation means for adoption economics, and what workspace admins should decide before enabling it. It is written for team leads and IT owners of Slack and Teams estates.

@ChatGPT: what OpenAI launched

The interaction is deliberately native: mention @ChatGPT the way you mention a coworker, in the channel where the discussion already lives. The assistant reads the request in context, can draw on authorized connected tools to actually do the work, and returns results into the thread where everyone sees them. Teammates contribute clarifications and corrections in-line, so steering the AI becomes part of the conversation rather than a private detour.

That placement matters because it removes the copy-paste tax that has defined workplace AI use: no lifting context out of Slack into a private ChatGPT session and carrying the answer back. The work request, its context, and its result stay in one shared, searchable place.

The no-subscription mechanic

OpenAI's phrasing — colleagues can participate without each needing an individual subscription — quietly changes the adoption math. In practice, one subscribed integration serves a channel full of people, some of whom have no ChatGPT plan. Every thread where @ChatGPT produces useful work is now a live demonstration running inside the company's own conversations, which is a sharper growth mechanism than any trial banner.

For budget owners it cuts the other way: value concentrates on fewer paid seats while exposure spreads. Expect the usage accounting — whose allowance a channel's requests draw from — to become a real administrative question as adoption grows.

How it fits the dots and Space architecture

@ChatGPT in Slack and Teams is one layer of a deliberate mesh from the same keynote. Dots — the always-on agents — are also reachable in Slack and Teams and carry context across channels; ChatGPT Space can ingest connected Slack context; and shared tasks can trigger on Slack messages. The chat platforms are being treated as the enterprise's nervous system, with OpenAI attaching at every junction: ambient assistant, persistent agent, workspace, and automation trigger.

The strategic positioning is notable given whose platforms these are. Microsoft Teams is the front door of OpenAI's largest partner's productivity suite; Slack belongs to Salesforce, an OpenAI Marketplace launch partner. OpenAI is distributing through the incumbents' own channels while building the surfaces — Space, Pages — that compete with what sits behind them.

Admin decisions before enabling

The integration's power comes from connected tools, which is where governance lives: an @ChatGPT that can act on authorized tools inside a public channel needs a clear map of which tools, authorized by whom, invocable from where. Data-handling defaults carry over — Business and Enterprise workspace content is not used for training by default — but admins should confirm how channel content, including messages from non-subscribers, is treated.

Rollout advice mirrors every chat-integration wave: start in a handful of active engineering or operations channels where requests are concrete, watch what people actually ask, and use that to shape tool authorizations before opening company-wide.

Dots overlap and platform response: what to watch

Watch three seams: how @ChatGPT and a team's dots divide labor in the same channels — the mention pattern for on-demand work, dots for standing responsibilities, with texting support for dots still to come; how Slack and Microsoft respond on their own assistant roadmaps now that a rival AI lives natively in their threads; and whether the no-seat participation model survives contact with heavy usage or acquires its own metering.

The adoption prediction is low-risk: mentioning an AI where the conversation already is beats switching apps, and the feature will spread wherever it is enabled. The governance work is deciding, before that spread, what it is allowed to touch.