Your guide to getting started with ZoomMate
Everything you need to start using ZoomMate: notes, search, workflows, and AI tools — all inside Zoom Workplace.
Select a language
8 min read
Published on October 26, 2026
AI agents are no longer experimental. They're in your meetings, summarizing customer conversations, drafting follow-ups, joining hiring workflows. They can touch real data, generate real content, and take real actions. Often without a clear audit trail.
But most organizations haven't closed the gap: they can't see what those agents are doing, what data they're accessing, or whether they're following internal policies.
That visibility gap is the hidden tax of AI adoption. It can scale with every new agent, every new workflow, every new integration. The faster you adopt AI, the wider it can get. Unless governance is built into the system itself.
At Zoomtopia, we announced five new AI governance capabilities for Zoom Workplace and our customer experience products, designed to give IT and security leaders the visibility and control they need to adopt AI confidently at scale.
What's new: Zoom recently introduced Centralized AI Management, custom AI guardrails, Deepfake risk detection, Sensitivity Labels (with optional Microsoft Purview integration), and Compliance Manager for My Notes and Zoom Canvas. Together, these capabilities are designed to give IT and security leaders visibility into AI agent activity across their organization, and control over what AI agents can access, generate, and share. Availability begins now and continues through January 2027.
"As organizations adopt AI across their operations, they need more than promises about security. They need tools that give them real visibility and real control," said Russell Dicker, Chief Product Officer at Zoom. "These capabilities are designed to help IT and security leaders govern AI activity with the same rigor they apply to every other part of their infrastructure."
| Capability | What it does | Availability |
|---|---|---|
| Centralized AI Management | Greater visibility into AI agent activity, account-level policies, AI activity audit controls | Expected Dec. 2026 |
| Custom AI guardrails | Rules for what AI can access and share, tailored to a customer's business-specific policies and requirements | Expected Q1 2027 |
| Deepfake risk detection | Real-time video/audio risk analysis, confidence indicator for synthetic media | Expected Dec. 2026 |
| Sensitivity Labels | Allows sensitivity labeling of Zoom Meetings and Whiteboard information within Zoom (with support for more products expected soon), with optional integration with Microsoft Purview | Available now |
| Compliance Manager support for My Notes + Zoom Canvas | Retention and archiving controls for AI-generated content | Expected Jan. 2027 |
Missed a session? Catch every Zoomtopia moment on demand
Centralized AI Management is an administrative capability available in Zoom Workplace that gives IT and security leaders a single view of AI agent activity across the organization. It includes policies set at the account level, audit controls, and the ability to manage AI agents created by employees, all from the admin console. It's expected to be available in December 2026.
Right now, most admins can't answer a straightforward question: "What did that AI agent just access?" As agentic workflows scale, that question gets more urgent, not less. AI agents are joining meetings, summarizing customer interactions, generating action items. In many organizations, nobody has a complete picture of which agents are running or what data they can reach.
Centralized AI Management is Zoom's control plane. One place to see what's running, enforce what's allowed, and audit what happened.
Administrators get policies at the account level that govern how AI agents behave, visibility into which agents are active and what data they can access, audit controls that track what AI accessed and shared, and oversight of agents that employees build on their own. When your CISO asks "what happened?", you have a real answer. When someone in marketing builds their own AI agent, you know about it.
Most organizations adopted AI capabilities one at a time. A meeting summary here, a chat assistant there. Nobody planned for who would govern the whole picture. Centralized AI Management is designed to close that gap before it compounds.
Custom AI guardrails are configurable rules in Zoom Workplace that let administrators control what information AI can include in its outputs. Admins can set rules to exclude specific topics or data from AI processing, and set limits on what data AI agents can output. They're expected to be available in Q1 2027.
If Centralized AI Management shows you what's happening, guardrails let you decide what's allowed to happen.
"Your data, your control" is the principle. In practice, that means an administrator decides what's in scope for AI, not the model. Guardrails helps govern what happens when AI moves from conversation to action to completion. They're the boundaries that make scaling possible without losing oversight.
An example: say you have a channel where your team discusses a sensitive new project. You probably don't want an AI agent summarizing those conversations and surfacing them in a search result. With Guardrails, an admin can exclude that channel entirely. Or require that any AI output touching that content goes through human review before anyone else sees it.
With Guardrails, admins can set rules to help limit what information AI surfaces in outputs, pause AI before it shares specific content types, exclude entire channels or meeting types from AI processing, and restrict what data AI agents can reach. The controls are designed to match how your organization actually works, not how a default setting thinks it should.
Centralized AI Management provides the visibility. Guardrails provide the rules. Together, they're the governance system for agentic workflows at enterprise scale.
Want to know more? Visit the Enterprise Trust web page
Deepfake risk detection is a trust and safety capability available as an add-on in Zoom Workplace that analyzes video and audio signals in Zoom Meetings in real time to help identify content that may be modified outside of Zoom. It flags potentially fabricated content and displays a confidence indicator to help participants assess authenticity and take appropriate action. It's expected to be available in December 2026 as a paid add-on.
You're interviewing a job candidate over Zoom. Their resume checks out. Their answers are polished. But something about the video feels slightly off. The lighting shifts when they move. The audio doesn't quite sync with the lip movements. You can't place it exactly.
Now multiply that scenario. A caller impersonating a trusted vendor to your contact center. A participant in a sensitive board meeting whose face and voice are digitally fabricated. These situations aren't hypothetical. Synthetic media is faster to produce and harder to spot than it was even a year ago.
Deepfake risk detection is designed to give people an early risk signal when something doesn't look right. During a live meeting, the system analyzes video and audio signals and, if it identifies patterns consistent with media modified outside of Zoom, it displays a confidence indicator. That visual cue helps participants decide whether to investigate further.
The capability is built for Zoom video meetings where verifying participant identity matters, hiring interviews where candidates may misrepresent themselves, and customer interactions where callers may attempt to impersonate trusted contacts.
Deepfake risk detection is designed to help identify potentially fabricated content and surface that information so humans can make the call. However, it won't catch everything. It's not a substitute for other verification methods. We're being upfront about that because trust starts with honesty about what a tool can and can't do.
Two additional capabilities help complete Zoom's AI governance approach. Sensitivity Labels (available now) allow organizations to classify data within Zoom Meetings and Zoom Whiteboard (more products coming soon) with labels that can be created natively in Zoom or optionally synced from Microsoft Purview, so labels follow the content wherever it goes. Compliance Manager for My Notes and Zoom Canvas (expected January 2027) extends retention and archiving controls to AI-generated content.
Governance controls are only as strong as the data protection underneath them. If a document classified as "confidential" in Microsoft 365 loses that classification the moment someone discusses it in a Zoom meeting, you have a gap, and it's exactly the kind of gap that surfaces during audits.
Sensitivity Labels helps solve this. Organizations can create sensitivity labels natively within Zoom, or optionally sync labels from Microsoft Purview. By syncing labels from Microsoft Purview, customers can apply the same sensitivity labels in Zoom that they use in Microsoft and support consistent policies across both platforms. The label travels with the content. Sensitivity Labels are included in Enterprise Plus and Premier accounts and can be purchased as an add-on to other licenses.
Compliance Manager support for My Notes and Zoom Canvas addresses a different problem. Zoom Compliance Manager already supports retention and archiving controls for meetings, phone, chat, contact center, whiteboard, and more. This announcement shares that Zoom will extend that coverage to AI-generated content in My Notes and Zoom Canvas, expected January 2027. Compliance Manager brings it under the same governance umbrella.
For organizations in industries where data retention and record keeping aren't optional, these two capabilities help close real gaps.
Zoom's approach to AI governance is built on a core principle: your organization should decide what its AI can access, what it can share, and how its activity is tracked, enabling administrators to keep control over policy enforcement across Zoom Workplace.
AI adoption is accelerating, and the organizations that move fastest will be the ones that trust their platform to handle governance at the system level, not the ones layering on oversight as an afterthought. This is what makes AI a system of action rather than a system of risk. Trust as the prerequisite for everything that follows.
We're not claiming this is the final word on AI governance. The regulatory environment is evolving. The way organizations use AI agents will keep changing. The threats we're building detection for today will evolve. But the foundation — visibility, control, data protection, and proactive threat detection — is what makes it possible to adapt as things change rather than scrambling to catch up.
Missed a session? Catch every Zoomtopia moment on demand
AI governance in Zoom Workplace is a set of administrative capabilities that are designed to give IT and security leaders better visibility into AI agent activity, control over what AI can access and share, and tools for data protection and compliance across the platform.
Centralized AI Management is an administrative capability in Zoom Workplace that gives IT and security leaders a single view of AI agent activity across the organization, with policies, audit controls, and management of agents created by employees.
Custom AI guardrails are configurable rules that help administrators control what information AI can include in its outputs, based on specific business policies and requirements.
Deepfake risk detection is a trust and safety capability that analyzes video and audio signals in real time and flags content that may be modified outside of Zoom, displaying a confidence indicator to help participants assess authenticity.
Sensitivity Label integration optionally syncs Microsoft Purview classification labels into Zoom Meetings and Zoom Whiteboard, helping to maintain consistent data labels across platforms without manual relabeling content.
No, Sensitivity Labels can also be created inside Zoom Workplace. Support for Purview is optional.
Compliance Manager extends retention and archiving controls to AI-generated and collaborative content in My Notes and Zoom Canvas, helping organizations meet certain compliance requirements.
Sensitivity Label with the optional integration with Microsoft Purview is available now. Centralized AI Management and Deepfake risk detection are expected in December 2026. Custom AI guardrails and Compliance Manager are expected in Q1 2027.
Yes. Deepfake risk detection is designed for use across Zoom video meetings, including hiring interviews and customer interactions.
Want to know more? Visit the Enterprise Trust web page
