Can Microsoft Purview See AI Usage on a Mac?

This question usually comes from someone with a mixed fleet and a Microsoft licence they are trying to get full value from. It is worth answering precisely, because both the optimistic and the pessimistic versions in circulation are wrong, and a security team that repeats either one loses credibility with the people reviewing the decision.

Can Microsoft Purview see AI usage on a Mac?

Partly. Purview Endpoint DLP, the data loss prevention agent that runs on the device itself, does support macOS, so any claim that Purview leaves Macs entirely uncovered is false. What it does not give you on macOS is the browser level visibility that the Purview DLP browser extension provides on Windows.

What exactly does the browser extension cover?

Microsoft documents the Purview DLP browser extension for Chrome and Firefox on Windows 10 and 11 (Source: Microsoft Learn, updated 15 June 2026). That is the component that inspects what happens inside a web page, which is where AI usage lives.

The more precise limitation, and the one worth putting in a board paper rather than the "Macs are blind" version, is narrower. Of the five site level actions Purview can take, three are available in Edge only. If your organisation has not standardised on Edge, the enforcement surface is smaller than the licence implies.

So where does the gap actually bite on a Mac?

Safari. It is the default browser on every Mac, it is what most executives and designers never change away from, and it sits outside the Purview browser extension entirely. An organisation running Endpoint DLP on macOS and assuming that covers browser based AI use is making an assumption the documentation does not support.

The second place it bites is the account. Endpoint DLP watches what leaves the device, but it does not tell you whether a person was signed into an AI tool with a corporate identity or a personal one. That distinction turns out to be most of the problem. Akamai measured 47.11% of enterprise AI conversations running through personal identities rather than corporate managed accounts (Source: Akamai, State of the Internet: Enterprise AI Usage Risk Report 2026, August 2026).

Is this a reason to replace Purview?

No, and any vendor telling you otherwise is overselling. Purview does a large amount of work that a browser control does not: information protection across the Microsoft estate, retention, eDiscovery, insider risk, and classification that follows a document rather than a session.

The right framing is placement, not replacement. Purview observes files and endpoints. Browser AI governance observes the interaction, what a person types, pastes and uploads inside a page. Those are different objects, and an organisation with obligations around both usually needs both.

Why can most security tools not see what is typed into a prompt?

Because each one observes a different object, and a prompt is not any of them. This is the part that settles the argument with a sceptical architect, because it does not require conceding that any existing control is bad.

• Application allowlisting observes processes. A browser tab is not a process.
• Firewalls and traffic platforms observe flows. A flow carries a destination, not an intent.
• Document DLP observes files. A pasted paragraph is never a file.
• TLS inspection observes payloads in transit, and only after the page has already assembled them.

Forrester found more than 70% of corporate work now happens inside a browser (Source: Forrester, Digital Workplace and Employee Technology Survey 2025). Governing at the point of intent means code running inside the page, which is a statement about placement rather than a claim about anyone's feature list.

How do I work out whether this gap matters to us?

Four questions, and they usually take an afternoon rather than a project.

1. What proportion of your fleet runs macOS, and what do those people actually browse in?
2. Have you standardised on Edge, or is that true only in the SOE document?
3. Would anything you own today distinguish a corporate AI account from a personal one?
4. Do contractors or partner staff reach your systems from devices you do not enrol?

If macOS is a rounding error, everyone genuinely uses Edge, and there are no unmanaged devices, your Purview licence is likely enough. That is a real outcome and we have delivered that answer to clients before.

What covers Safari and the account question?

A control that installs into the browsers themselves. Akamai Workforce Protector is what we deploy for this, through an extension covering Chrome, Edge, Firefox and Safari, with an optional endpoint agent when coverage needs to reach desktop AI applications and AI enabled IDEs.

Be precise about that boundary when you present it internally, because an architect will test it. Browser coverage is agentless and deploys through the device management you already run, with no proxy and no traffic redirection. Desktop applications and IDEs are not agentless and do need the endpoint agent.

One detail matters for an Australian privacy review. Classification happens inside the browser, and Akamai's position is that content, including personal information, stays there while only alerts reach the cloud console. That is an architectural answer rather than a contractual undertaking, which is a materially different thing to put in front of a privacy officer.

How does David and Goliath help?

We audit what your current licences genuinely cover before recommending anything, because organisations routinely buy a second product to solve something the first already handles. Where Purview covers it, we say so.

Where it does not, we size the gap with evidence rather than assertion: which browsers your staff use, which populations sit outside managed policy, and what is reaching AI tools through personal accounts today. The method is in how to find out what AI tools your staff are actually using, and the equivalent analysis for Google's stack is in does Chrome Enterprise Premium cover Safari.