Copilot with commercial data protection only sees the public web

One question keeps landing in my inbox lately. If Copilot with commercial data protection is free for so many Microsoft 365 users, why would anyone pay for Copilot for Microsoft 365? The confusion is real, and I get it. The names are close, and Microsoft has been pushing AI tools into its products at a fast pace. In several tenants I work with, both users and IT mix the two up. They don't see the differences in data access and privacy, and above all they don't know when that green "Protected" badge actually matters.

Copilot with commercial data protection only draws on public web data

In November 2023, "Bing Chat Enterprise" was renamed "Copilot with commercial data protection". Since then it has been included at no extra charge for users with eligible work accounts (Entra ID) on Microsoft 365 Business Standard, E3, E5 and other plans. Microsoft promises that prompts and responses are not saved, not used to train the models, and that nobody at Microsoft has "eyes-on" access to the conversation. When this mode is active, you'll see a green shield marked “Protected” in the Edge sidebar, at copilot.microsoft.com or in Windows 11 (23H2 and later). There is one big limitation, though. This Copilot uses public web data as its only source. It can't see or reference documents, emails, Teams chats or files in your Microsoft 365 tenant, and that boundary is the whole point.

Many of the organizations I advise are still arguing about whether to allow this free Copilot or block it from the start. My take is that it's a very low-risk way to let staff get real experience with AI without putting any internal data on the line. If you want Microsoft's own documentation, read the Copilot overview and the page on managing Copilot.

Copilot for Microsoft 365 costs extra but works with your tenant data

Copilot for Microsoft 365 is a completely different product. With the paid add-on license it works with Microsoft Graph, so it can reach your emails, calendar, Teams messages, SharePoint, OneDrive and more, always in the user's security context. That's where its value comes from. It's also why security, privacy and governance teams start asking hard questions. If permissions, sharing and access controls in your tenant aren't in order, sensitive information can surface where it shouldn't. I covered what that involves in an earlier post, Copilot for Microsoft 365 tenant readiness.

The full Microsoft 365 Copilot is never free, unlike Copilot with commercial data protection. So the question to ask goes beyond "Who gets Copilot?". You also need to work out how people end up using the right Copilot, and using it securely.

The green shield and the work account decide what users get

This is where day-to-day use gets messy. Plenty of users simply search for "Copilot" or "Bing Chat" and have no idea which account they're signed in with. The green "Protected" badge isn't decoration. It's the only way a user can tell that commercial data protection is on. Sign in with a personal Microsoft account and the shield disappears, and so do the privacy guarantees for prompts and responses.

My tip: Spell it out in your user communication. Staff should always look for the green shield and sign in with their work account. A screenshot and a one-page note on the intranet will do more than a 20-page AI policy.

Admins decide whether and how Copilot with commercial data protection is available. You manage it in the Microsoft 365 admin center, in Entra ID (formerly Azure AD) and through Edge policies, and you can turn it off for specific users or groups if you need to. Microsoft's admin guidance is at Copilot management.

When I explain sign-in and protection to customers, I usually draw this diagram:

graph TD
    User["User"] --> SignIn["Sign in with work account"]
    SignIn --> Check["License & Entra ID checked"]
    Check --> Protected["Green shield – Protected"]
    Protected --> CopilotWeb["Copilot web answer"]
    Check --> NoProtection["No protection with personal account"]
    NoProtection --> CopilotWeb

Figure: Sign-in and protection mode in Copilot with commercial data protection.

The free Copilot gives you room to shape your AI policies

Copilot with commercial data protection has no view of tenant data, which makes it a good place to start hands-on AI trials and work on internal policies. I often come across companies that haven't taken any position on AI tools, so their staff use ChatGPT or similar services with no oversight at all. Turning on Copilot with commercial data protection lets people experiment in a setup that is "safe by default", with no risk of data leaking out of your tenant.

It doesn't replace a strategy, of course. If Copilot for Microsoft 365 is on your roadmap, start working on permissions, oversharing and transparency now. My post Copilot for Microsoft 365 tenant readiness walks through the practical steps.

Clear communication matters more than technical complexity

I see Copilot with commercial data protection as the ideal entry point for AI in an organization. The distinction is easy to remember. The free Copilot works with web data only, and it's protected when the shield shows. The paid Copilot for Microsoft 365 has full access to your internal data. Explain the shield clearly and make work account sign-in the rule, and you get the benefits of AI at very little risk, long before you have to deal with the full privacy implications of Copilot for Microsoft 365. When you do take that step, your users and your data governance will be in much better shape for it.