Microsoft is giving Teams meeting organizers something they’ve been asking for: a genuine off switch for AI.

Starting this month, licensed organizers and presenters can toggle Meeting AI on or off directly during a live call. The control covers Copilot, Facilitator, and Recap, the three AI-powered features Microsoft has been rolling into every Teams meeting over the past year. When the toggle goes off, none of those systems generate responses, notes, or summaries for the rest of that session.

The rollout started with Targeted Release users in early July and is expected to reach general availability by the end of the month, covering Windows, macOS, mobile, and web.

What Prompted the Change

The immediate trigger was Teams Facilitator, a feature Microsoft introduced earlier in 2026 that listens to meetings in real time and offers proactive responses before anyone asks a question. The intent was to make AI feel ambient and always ready. The result was a wave of discomfort from employees and IT administrators who felt the feature crossed a line between assistant and surveillance.

The response was swift enough that Microsoft moved to give organizers an override. It is worth noting the context: this is not a privacy setting that employees can control themselves. The toggle belongs to the meeting organizer or presenter. Individual attendees cannot turn the AI off on their own.

How the Controls Work

When Meeting AI is disabled, Teams will not generate Copilot responses, Facilitator answers, or automated meeting notes for the remainder of that session. The change does not apply retroactively to previous meetings or recordings.

Organizers can also choose to disable features individually rather than shutting everything off at once. Copilot and Recap can be toggled independently while keeping Facilitator active, or the other way around.

There is one important dependency: transcription and Meeting AI are linked. Turning on Meeting AI automatically enables transcription and Recap generation. Similarly, starting a transcription session activates Meeting AI and Recap. They cannot currently be separated.

The controls are available to anyone in the organizer or presenter role using Teams on the web or desktop.

Why This Matters for Enterprise Teams

This rollout reflects a broader shift in how enterprise AI tools are being deployed in 2026. The early wave of AI feature introductions assumed workers would embrace ambient assistance. Many did not, particularly in contexts where the meeting involved sensitive client discussions, confidential business decisions, or simply situations where people wanted to speak candidly without a transcript being generated.

The pushback against Teams Facilitator is one of the clearest examples of what happens when AI features are turned on by default without sufficient consent architecture. Microsoft’s response, adding granular controls, is a reasonable first step, but the fact that individual participants cannot override the organizer’s decision raises ongoing questions about whose interests the default settings serve.

For businesses running Microsoft 365 Copilot licenses, the new controls add a useful layer of governance. Compliance teams in regulated industries, law firms, healthcare organisations, and finance teams dealing with material non-public information now have a formal mechanism to prevent AI recording and summarisation in sensitive sessions.

What This Means for Business

If your team uses Microsoft Teams with Copilot licenses, this update deserves attention from both your IT administrators and your leadership team.

The practical questions are straightforward. Which meeting types should have AI disabled by default? Who in your organisation controls the organiser role for external client calls? Are there meeting categories where transcription creates compliance risk?

The answers will differ by industry. A sales team running internal pipeline reviews might want full Copilot access. The same team running a call with a prospect mid-negotiation might not. The new toggle makes those policies enforceable rather than relying on individual decisions in the moment.

The deeper issue is consent architecture. As AI becomes more capable of generating real-time meeting intelligence, the governance question shifts from whether to use these tools to who gets to decide when they are active. Microsoft’s move to give organizers that control is progress, but it also reinforces that this governance conversation needs to happen at the business level before those meetings start.

Organisations that have not yet established AI-in-meetings policies should treat this update as a prompt to do so. The tools are now granular enough that a policy is both practical and necessary.

---

Enterprise DNA works with business leaders building AI strategies that balance productivity with governance. If your team is navigating how to deploy AI tools like Copilot in a compliant, consent-aware way, we can help.