ChatGPT now available via @mention in Slack and Microsoft Teams

At 20:13:58 UTC on September 29, 2026, the official ChatGPT X account posted that users can now @ChatGPT in Slack and Microsoft Teams. The post shows 7,806 views, 13 replies, 1 repost, 3 quotes, 6 likes, and 6 bookmarks.

In product terms, this is not a standalone app but an embedding of ChatGPT into two existing team communication platforms. The official description of how to use it is straightforward: @ChatGPT in a channel, thread, or DM to invoke the feature.

It should be noted that all the feature details above come from the OpenAI official account's own claims, and no third-party verification appears in the source. So this article's description of these capabilities should be understood as the official account's position, not as independently confirmed fact.

The core scenario given officially is turning messy threads into plans, slide decks, or spreadsheets. In other words, the input is team conversation, and the output is a ready-to-use structured document.

Another emphasized capability is context expansion: ChatGPT can pull in context from beyond the thread via authorized connected sources. That means it is not limited to the current conversation content, but only on the premise that authorized connected sources are used.

The company also mentioned a collaborative aspect: team members can add context and improve results within the same conversation. This makes ChatGPT's output not a one-off generation but something multiple people can continue to revise.

The company explicitly stated that using the feature requires no individual ChatGPT license. This lowers the barrier for individual employees—there is no need to first open a ChatGPT subscription for themselves.

But the feature is not for all users: it is offered for Business and Enterprise plans. In other words, the barrier shifts from 'individual license' to 'organizational plan.'

These two constraints need to be understood together: if an organization has already purchased the relevant plan, individual members can use it without holding a license themselves; conversely, for users outside that plan range, the official account gives no available path.

The company did not specify the feature's specific available regions or phased rollout timing. For an integration spanning both Slack and Microsoft Teams, this directly affects when users can actually use it.

The company also did not specify which specific connected sources are supported. Since 'pulling in context from beyond the thread' depends on authorized connected sources, the range of available sources is a key variable in the feature's boundaries, but the official account does not provide it.

Also unspecified: whether the feature has been fully rolled out to all Business and Enterprise users, and whether its capabilities differ between Slack and Teams.

In replies to the official post, a user said they had not yet seen the option. This is a confirmed piece of user feedback, but by itself it cannot prove the feature is not live, nor can it indicate how large the affected scope is.

Putting this together with the company's failure to specify rollout timing or whether it is fully rolled out, the more reasonable reading is that there may be a time gap between the announcement and full availability. This is an inference based on the source information, not official confirmation.

So the accurate understanding of this news is: OpenAI announced the feature and its scope of applicability, but real-world availability remains uncertain.

The only source for this information is the post from the official ChatGPT X account, which is an official self-claim. There is no third-party verification in the source, nor independent testing or user-scale data.

The post's own engagement data—7,806 views, 13 replies, 1 repost, 3 quotes, 6 likes, 6 bookmarks—can only indicate the announcement's reach, not judge feature quality or actual adoption.

Under this evidence structure, the feature's existence and positioning can be recorded per the official account, but judgments about effectiveness, stability, and coverage should be held back.