About this step

Step 2 of the AI Security Setup wizard is where you choose which DLP policies apply to AI interactions — the rules that detect and protect sensitive data flowing through the apps you connected in Step 1.

The page title asks "What data should AI Security monitor?" and the subtitle reads "Use ARIA to get tailored recommendations, or select policies manually." The catalogue contains 330 predefined policies across six categories (Identity & PII, Financial, Healthcare, Technical & IP, Legal & Compliance, HR & Workforce), and you must enable at least one policy — or click Skip For Now — before the Next button activates.

You have four valid approaches:

Approach How-to Best when
ARIA guided questionnaire 1 You want tailored recommendations and don't mind 3 questions
ARIA free-form chat 2 You already know your context and want to skip ahead
One-click re-use of existing DLP 3 Your tenant already has Cloud DLP configured
Manual catalogue browsing 4 You know exactly which policies you want

Two supporting how-tos:

# How-to When to use it
5 Inspect and fine-tune a policy's detection rules You want to see what a policy actually detects or toggle specific rules
6 Skip Step 2 and configure policies later You want to finish setup now and return to policies after