Separate text retention from evidence retention
Teams may need short-lived processing for submitted text while keeping run IDs, risk bands, reviewer notes, and policy status for auditability. Treat those records as separate retention classes.
Resources
Plan AI detection API retention rules for submitted text, scan results, reviewer notes, audit logs, privacy reviews, and policy-sensitive writing workflows.
Open core guideTeams may need short-lived processing for submitted text while keeping run IDs, risk bands, reviewer notes, and policy status for auditability. Treat those records as separate retention classes.
Education, publishing, HR, and enterprise workflows should define who can access text, how long raw content is stored, when results expire, and which roles can export records.
Retention planning should include deletion windows, customer export needs, reviewer history, and legal or institutional requirements so teams do not improvise after a dispute.
Most teams should distinguish raw submitted text, extracted evidence, scores, reviewer notes, run identifiers, audit events, and account metadata, then set a retention period for each class.
No. Sensitive writing should not be retained indefinitely by default. Teams should choose short retention for raw text unless policy, audit, or customer requirements justify longer storage.