Privacy policy

Short version: this website has no accounts, no cookies, no analytics, and no forms. When you run Darwin yourself, your data stays on your machine and goes only to the model and sandbox providers you configure.

This website

When you run Darwin

Darwin is self-hosted. Your tasks, genomes, and run records are written to your own disk. If you enable the optional integrations, prompts and code are sent to the providers you configure (Fireworks AI, Braintrust, Daytona) using your own API keys and under their terms. With --offline, nothing leaves your machine.

Changes and questions

Changes to this policy are tracked in the public repository history. Questions go to karthik.subramanian@berkeley.edu.