Create a policy

Policies are the rules Shield enforces in the browser: what may happen, with which data, on which AI tools. You can start from a compliance preset — or build your own in a couple of minutes.

Create the policy

1. Open Policies and select New policy.
2. Choose what it applies to: All tools, or Specific tools picked from the AI tools discovered in your organization. Tool-specific policies stack on top of general ones — on conflict, the stricter action wins.
3. Select the data types the policy covers — for example PII, financial data, source code, or secrets and credentials. See data types & detection.
4. Select the actions it governs: submit prompt, upload file, copy/paste, download, share.
5. Pick the enforcement mode — from observe-only to a full block. See enforcement modes explained.
6. Name and description are pre-filled from your selection; adjust them and select Create policy.

For a tool you want gone entirely, use Block entire tool — it blocks all use of that tool, and data types and actions no longer matter.

Manage the lifecycle

Policies are Active or Paused — pause instead of deleting when you want to try something out, and bulk-activate or pause from the list. Every edit increments the policy's version number, so audits can trace exactly which revision was in force.

What your team sees

Active policies show up for members in the Shield Web side panel under Policies, and enforcement happens directly in the browser on the next page load of the affected AI tools.