Let your employees use AI. NativeAI Guard is the safety net.

We complement it. Purview protects the structured channels it was built for: mail, endpoints, SharePoint. Its shadow-AI DLP runs only in Edge for Business, billed by usage and driven by static rules, and static rules and classification labels miss content whose sensitivity lies in its meaning, and the AI surface beyond the Microsoft perimeter. A Swiss bank's head of IT security put it plainly: closing that remainder inside Purview costs six figures in consulting; buying the layer is cheaper than building it. NativeAI Guard adds digital sovereignty: no new dependency on a US company.

Today: Chrome and Edge through the browser extensions, covering every browser-based AI tool, translation service and webmail, plus mobile devices; a drop-in API proxy protects the AI applications you build yourself. In active development: a system-level endpoint agent for desktop applications and command-line tools such as Claude Code. On the roadmap: macOS, Firefox and Safari, in the order the organizations working with us actually need.

Transient processing is the standard mode: the prompt is inspected in volatile memory on our own servers in Switzerland and discarded immediately, nothing of its content written to disk. What persists is the audit entry: who sent something, when, to which model, and what NativeAI Guard did about it. Retaining prompt content is an option for customers with record-keeping duties, banks in particular, and is switched on deliberately.

No operational data. Prompts, metadata, decisions and usage logs generated through your use of the platform are never used to train, tune or validate machine-learning models, and that is written into the data processing agreement rather than left as a policy statement. Our classification models are pre-trained on synthetic and public corpora. The one exception is explicit and consent-based: if you choose to provide an anonymized calibration dataset during onboarding to tune detection to your own documents, that use is named in the contract and happens only because you asked for it.

Both, and the better answer is often a third option: anonymization. NativeAI Guard replaces the sensitive elements (names, client numbers, patient identifiers, contract references) before the prompt reaches the model, so the employee still gets a useful answer and keeps working. Hard blocking stays reserved for categories where no exception is acceptable, and every block is logged with the policy violation in the audit log.

Every domain can have its own rule, and a default rule applies automatically to all domains until you adjust it for a specific one. Each rule runs in its own mode: anonymization, policy enforcement, allow or block, so the same deployment can be strict in one place and light-touch in another. You write new rules in plain language by typing them in, and you adjust existing rules the same way. You create and manage user groups yourself, so rules can be enforced differently per team or department. The AI governance dashboard shows you at any time which rules are active and how they are being applied.

Password-protected files are intercepted at upload: the user enters the password and the content is scanned in the clear before anything is sent. Documents marked confidential are blocked outright; non-confidential ones are content-inspected, so a file whose label was removed is still caught. Rights-management-encrypted files (Microsoft Purview Information Protection, Azure RMS) are not inspected yet; until that ships, a policy can stop them at upload, so an uninspectable file is a decision, not a gap.