Auto-flag PII and SPII

What's new

Outset now flags sensitive personal details ("SPII" like health, race or ethnicity, religion, sexual orientation, biometrics) alongside personal identifiers (PII) like names, addresses, and account numbers. Detection covers spoken and typed answers, participant uploads, and information shown on video or screen share.

Why it matters

Participants share sensitive information whether or not you asked. Answers with unresolved flags stay out of results, downloads, analysis, exports, chat with your data, the API, and MCP until a reviewer keeps or deletes them.

Used for

• Healthcare studies where participants describe conditions or treatments

• Financial services research where account details come up unprompted

• Screen-share studies, where information appears on screen without being spoken

• Organizations that need one privacy rule across every study

How to use it

Org admins set the policy. Turn on Auto-flag PII, Auto-flag SPII, or both, and choose whether reviewers can see flagged content or only delete it. What you set here applies to every study and cannot be changed lower down.

Researchers configure it per study when the organization has not enforced a category.

Whoever owns review clears the flags before results and downloads unlock, keeping or deleting each item or handling them in bulk. If auto-flagging misses something, they can flag it from the transcript themselves.

Read more about how it works in the Help Center