Skip to main content
Genie scans every message sent in workspace chat and project chat against a set of PII detection rules before it is processed. Workspace owners and admins can configure which types of sensitive data to detect, choose what happens when a rule matches, and create custom rules for data that is specific to their organisation.
PII rules are available on the Teams plan. Only owners and admins can configure rules. All workspace members can trigger detections.

​
Open PII settings

In the workspace sidebar, click Settings, then click PII Detection — or navigate to Settings → PII Detection from the settings page. The page has two sections:
  • Detection rules — built-in detectors and any custom rules you have created, each with a toggle and an action dropdown.
  • Detection log — a read-only history of recent matches. Matched values are always shown censored; the raw content is never stored.

​
Built-in detectors

Genie ships with detectors for common categories of sensitive data. All are enabled by default with the log action.
DetectorWhat it catches
Email addressEmail addresses in any format
Credit card numberMajor card formats (Visa, Mastercard, Amex, and more)
IBANInternational bank account numbers
API keyGeneric API key and secret patterns (Bearer tokens, hex secrets)
Phone numberInternational and local phone formats
Social security numberUS SSN formats
Passport numberCommon international passport formats
IP addressIPv4 and IPv6 addresses
You can change the action on any built-in detector or disable it. You cannot delete a built-in detector — disable it instead, or use Reset to defaults to restore every built-in to its shipped enabled state and log action.

​
Custom rules

Create custom rules for data patterns specific to your organisation — for example, employee IDs, contract numbers, or proprietary reference codes.
1

Click Add rule

On the PII Detection page, click Add rule.
2

Name the rule

Give it a clear display name (for example, “Employee ID”). This name appears in the detection log when the rule matches.
3

Enter a regex pattern

Write a regular expression that matches the data you want to detect. The pattern is validated before saving — an invalid regex returns a clear error message. Example: \bEMP-\d{6}\b catches employee IDs like EMP-123456.
4

Set the confidence level

Choose Definite (the pattern unambiguously identifies sensitive data) or Possible (the pattern may contain sensitive data). This label appears in the detection log.
5

Choose an action

Select what happens when the rule matches. See Actions below.
6

Save

Click Save rule. The rule becomes active immediately for all workspace members.
To edit or delete a custom rule, use the action menu on the rule row.

​
Actions

Each rule has an action that controls what happens when it matches a message.
The message is allowed through. A censored entry is written to the detection log — the matched value is stored as [REDACTED] and the raw content is never recorded.Use this for: data you want to monitor without disrupting users.

​
Detection log

The Detection log section shows the 50 most recent rule matches. Each entry includes:
  • The rule that matched (e.g. “Credit card number”)
  • The action taken (log or block)
  • The chat surface where the match occurred (workspace chat or a project)
  • The workspace member who sent the message
  • The date and time
Matched values are always shown as [REDACTED]. Only owners and admins can view the log.

​
Manage rules from workspace chat

You can read and update PII rules by asking the Genie workspace AI in plain language:
  • “List all my PII rules” — shows you the full rule list with their current actions.
  • “Block credit card numbers instead of just logging them” — the AI updates the rule’s action to block.
  • “Add a rule that catches employee IDs matching EMP-123456” — the AI creates a custom rule on your behalf.
  • “Show me the most recent PII detections” — the AI reads the detection log and summarises what was caught.
Mutations made via workspace chat require the same owner or admin role as the settings UI, and the AI will ask for confirmation before weakening a block action or disabling a detector.