Zero data retention.

What gets deleted?

A recurring cleanup job removes run content: the task, the agent's steps, screenshots, output, files and logs. A finished agent run is normally cleaned up shortly after it ends. Stored files are deleted, and the content of database records is wiped.

v3 sessions go further: results are never written to our database and reach you only through your webhook.

What do you give up?

Most features whose job is to remember are off on a zero data retention project:

• Browser Use profiles. They are off by default. You cannot create a profile or attach one to a session, so sessions start with a clean browser and no saved logins. If a workflow needs to sign in, it signs in each time.
• Workspaces and file uploads. Files do not persist between runs.
• Stored secrets and 1Password vaults for agent runs, integrations, and Stripe Link.
• Skills, schedules and share links.
• Session recordings of agent runs.
• Team members and the dashboard chat. A zero data retention project is used through the API, and the dashboard shows an API-only page instead of the chat.
• Some models. Models whose provider keeps request data are refused.

What does it not mean?

• Nothing is instant. Data exists while the run is working and until the cleanup that follows it. Cleanup can take longer than one sweep, for example while a result is still being delivered or when a purge fails and is retried.
• Your webhooks still fire. That is how results reach you.
• The model still sees the task. Prompts go to the model provider to be processed, the same as on any project.
• It is not HIPAA mode. HIPAA mode adds model-routing and data-handling restrictions; zero data retention controls how long we keep run content.

Getting it

Zero data retention is part of an enterprise agreement. Email sales@browser-use.com and we will switch it on for your project.