Data loss prevention for AI tools

Data loss prevention (DLP) for AI is the set of controls that stop sensitive information leaving your organisation through an AI tool — whether sanctioned or not. The discipline isn’t new; what’s new is the volume of low-friction ways to paste data into something external. Controls that assumed file transfers and email need to account for a chat box.

Start with what you’re protecting

DLP fails when it tries to protect everything equally. Classify first: what data actually matters — customer records, credentials, unreleased financials, regulated personal data — and concentrate effort there. A control that blocks everything gets switched off or worked around, which is the shadow-AI cycle again.

Layers that hold up

• Classification and labelling. Sensitivity labels that travel with the data let downstream tools enforce policy instead of guessing.
• Tenant-level policy. For sanctioned tools like Copilot, configure DLP so labelled content isn’t summarised, surfaced, or sent to connectors that shouldn’t see it.
• Egress controls. Network and endpoint DLP can flag or block sensitive content heading to known consumer-AI endpoints. Imperfect, but it raises the floor.
• A sanctioned alternative. Egress controls without a good internal option just push staff to creative workarounds. Pair enforcement with a private tool — managed Copilot or a local model.
• Logging and review. You can’t improve what you can’t see. Audit trails on the sanctioned tools turn incidents into lessons.

Good DLP is a ratchet, not a wall: classify what matters, give people a safe route, and make the unsafe route visible and inconvenient.

Treat it as ongoing

AI tools, plugins, and endpoints change monthly. A DLP posture set once and forgotten drifts out of date fast. Review the list of sanctioned tools, connectors, and blocked endpoints on a schedule, not only after an incident.