Security and trust
A plain-language summary of how we protect your work and your data.
Approvals
Playbooks stop for a named person to approve before anything is sent or saved outside. The person who started the work can't approve it themselves, except in one-person organizations, where it is labelled self-approved.
Sources you can check
Drafts point to the exact passage in your documents. Claims that can't be matched to a source are removed and flagged.
Sensitive data blocking
Card numbers, bank details, government ID numbers, health information, and student records are blocked before any AI sees them.
Your data stays separate
Each organization's data is kept apart at the database level, and your documents are never used to train AI.
Encryption and backups
Data is encrypted in transit and at rest, and backed up.
A complete record
Every run, approval, and change is written to a tamper-evident audit log you can export.
Who helps us run the service
The companies that process data for us are listed on the subprocessors page.
See the subprocessorsSecurity contact
To report a security problem, email the address below.