Keep the evidence. Release the raw source.
Retention depends on the purpose of each data class. Shield minimizes durable source content while keeping the evidence, decisions, and security records needed to make the workflow reviewable.
Process the source transiently.
Conversations, emails and documents are scanned on your device, and a report is built from the counts the device sends: categories, severities, the length and the source type. No source text or file reaches Shield to be stored. Reports made earlier from uploaded content keep masked findings and hashes until they expire.
Expire reports on a defined schedule.
An evidence report expires 90 days after it is created, on every plan. A Firestore TTL policy deletes it automatically rather than on request, while audit events preserve the minimum operational trail.
Use short-lived inputs and outputs.
The file cleaner runs in your browser: the original, your selections and the cleaned copy stay on your device. Shield records a receipt of each file as an audit entry: its type, how many findings of which categories, whether OCR was used and the engine version, and once the copy is saved, the change counts, the approval and any warning codes. Jobs from the earlier cloud cleaner kept a private uploaded original, encrypted selections, the cleaned output, the job status and verification data; the originals were deleted two days after upload and the outputs are deleted thirty days after they were made, through expiry rules rather than public permanent URLs.
What they stored was deleted on September 11, 2026.
Breach and dark web monitoring, the open-web exposure scan, and privacy requests were retired on September 10, 2026, and no new lookups, searches, or requests are made. For breach results Forg3t kept normalized source, date, category, severity, counts, and hashes; it never requested or retained raw passwords, extracted breach values, or third-party provider credentials from the user. All of it, the breach and exposure results, the identity vaults, and the privacy-request records with their timelines, was deleted on September 11, 2026. A copy exported just before is kept in a private storage bucket in the EU for 30 days in case a deletion has to be undone, and is then deleted automatically.
Choose how long your own history stays.
Scan history and finding decisions are kept for the number of days you choose in Account settings, from 1 to 365, 30 unless you change it. An organization's owner or an admin chooses how long its own records stay, its daily department totals and its approval requests, in the professional workspace, from 30 to 3650 days, 365 unless they change it. A job runs every day and deletes what is older, including what was stored before the setting changed. Evidence reports stay 90 days on every plan, whatever the setting.
Deleted is gone after a short, fixed window.
A stored file that is deleted stays recoverable by Forg3t for seven days (soft delete), then it is gone. The database keeps earlier versions of its records, deleted ones included, for seven days of point-in-time recovery, and a daily backup of the whole database is kept for thirty days. Only Forg3t can read either, to restore the service after a failure.
Revoke access and orchestrate deletion.
Account deletion revokes refresh access and runs retryable cleanup across account data, storage objects, credentials, billing links, and identity records.
Ask, or delete the account.
Write to privacy@forg3t.io to delete a stored record before its retention period ends, or delete your account in Account settings to remove everything at once.
