Creating DLP Policies

Getting Started with DLP Policies

A DLP policy defines what to detect and what to do when sensitive data is found. You can create multiple policies for different data types or use cases.

Each policy can have multiple detection patterns and a specific action (block, redact, alert, or log-only).

Creating Your First DLP Policy

1 Navigate to AI DLP

From your dashboard, click AI DLP in the sidebar, then click the Create Policy button.

2 Name Your Policy

Give your policy a descriptive name that explains what it protects. For example:

"Block SSN and Credit Cards"
"Redact Customer PII"
"Alert on API Key Exposure"

3 Select Detection Patterns

Choose from pre-built patterns or create custom ones:

PII Patterns - SSN, passport, driver's license
Financial - Credit cards, bank accounts
Credentials - API keys, passwords, tokens
Custom Regex - Your own patterns

4 Choose an Action

Select what happens when sensitive data is detected:

Block	Prevent the message from being sent to the AI
Redact	Replace sensitive data with [REDACTED] and allow
Alert	Allow but send immediate notification
Log	Allow and log silently for review

5 Save and Activate

Click Create Policy to save. Your policy is immediately active and will start scanning traffic.

Understanding Policy Actions

Block

The most restrictive action. When sensitive data is detected, the entire message is blocked from reaching the AI. The user sees an error message explaining why their request was blocked.

Use Block for highly sensitive data like SSNs, credit cards, or credentials that should never be sent to AI systems.

Redact

Replaces detected sensitive data with [REDACTED] or *** before sending to the AI. The user's request still goes through, but without the sensitive information.

Example:

Original: "Customer John Smith (SSN: 123-45-6789) called..."
Redacted: "Customer [REDACTED] (SSN: [REDACTED]) called..."

Alert

Allows the request but sends an immediate notification to administrators. Use this when you want visibility without blocking legitimate workflows.

Log Only

Records the detection silently without alerting or blocking. Useful for auditing and understanding data flow before enforcing stricter policies.

Testing Your Policy

After creating a policy, test it to ensure it's working correctly:

1 Go to Your Policy

From the AI DLP page, click on your policy to view its details.

2 Use the Test Panel

Enter sample text containing sensitive data patterns. Click Test Policy to see what would be detected.

3 Review Results

The test shows exactly what patterns matched and what action would be taken. Adjust your patterns if needed.

Always test with fake/sample data, never real sensitive information. Use test SSNs like 000-00-0000 or test credit card numbers.

Once you're confident your policy works correctly, it will automatically protect all AI traffic that passes through Signal Canary.