Microsoft is rolling out general availability of a “Meeting AI” toggle in Teams through late July 2026, giving licensed organisers and presenters a live, in-meeting control to switch off Copilot, Facilitator, and Intelligent Recap — either individually or all at once. It’s a direct response to months of user discomfort with AI features that were effectively always-on by default, listening in and generating transcripts, summaries, and notes whether or not participants had actively opted in. For UK SMEs running client calls, internal reviews, or anything sensitive on Teams, this is a small interface change with real practical consequences for what gets recorded and where that data goes.
Why this matters beyond convenience
The backlash Microsoft is responding to wasn’t just about preference — it was about consent and control. When meeting AI runs by default, every participant’s words potentially get transcribed, summarised, and stored, including in meetings involving legal advice, HR conversations, client negotiations, or anything a business would reasonably want to stay off the record. Not every attendee necessarily knows AI features are active, and not every organisation has thought through where those transcripts and summaries live afterwards, who can access them, or how long they’re retained. The new toggle puts that decision explicitly in the organiser’s hands during the meeting itself, rather than buried in an admin settings panel nobody checks.
What to actually do with the new control
Don’t just note that the toggle exists — decide as a business when it should and shouldn’t be used, and make that decision before your team has to guess in the moment. A sensible starting point: default meeting AI to on for routine internal meetings where a summary genuinely saves time, and build a habit of switching it off for anything involving client-sensitive discussion, HR matters, or legal conversations. If you run client meetings regularly, it’s worth telling clients upfront whether AI note-taking is active — being asked afterwards “was that meeting being transcribed?” is a worse conversation than mentioning it at the start. This sits alongside the broader GDPR question around AI meeting tools generally: recorded transcripts of identifiable individuals count as personal data, and having a clear internal policy on when AI capture is appropriate is worth doing regardless of this specific Microsoft update.
If your business runs on Microsoft 365 and you haven’t reviewed your AI feature settings across Teams, Copilot, and related tools in the last few months, this is a reasonable trigger to do that properly rather than leaving defaults as Microsoft ships them. ApplyAI works with UK SMEs specifically on setting sensible AI governance and usage policies for tools like this, rather than just switching things on and hoping it works out.
What IT admins should check separately
The organiser-level toggle is only half the picture. Microsoft 365 admins can also set tenant-wide and policy-level defaults for meeting AI, controlling whether it’s on by default for the whole organisation, restricted to certain licence types, or off unless explicitly enabled. For most SMEs without a dedicated IT function, nobody has actually looked at this setting since Copilot was first switched on — it’s usually whatever the default was at rollout. It’s worth five minutes in the Teams admin centre confirming what your organisation-wide default actually is, because the organiser toggle only helps if people know it’s there and remember to use it in the moment. Setting a sensible tenant default removes that reliance on individual judgement during a live call.
The takeaway
Once the Meeting AI toggle reaches your tenant, agree internally on when it should be used and when it shouldn’t, tell clients when it’s active for calls involving them, and treat this as a prompt to check your wider AI settings across Microsoft 365 rather than a one-off UI update to ignore.
