Trust Center

Built for organisations that take trust seriously.

Everything you need to evaluate Boxes — security posture, compliance, privacy, subprocessors, AI governance. Reports available under NDA from security@boxes.sh.

AI governance commitments

  • We do not train models on customer content. This is contractually committed in our ToS and DPA.
  • Every agent action is cryptographically signed. Tamper-evident audit log with five-year retention.
  • Capability scopes are enforced server-side. Agents cannot exceed their granted permissions.
  • Human approval gates are configurable. Default to "human-in-the-loop" for any new agent capability.
  • Customer-controlled model routing. Bring your own LLM. Disable third-party inference per workspace.
  • Pre-action visibility. Every draft viewable before send, with confidence and rationale.

Need something specific?

Audit reports, custom DPAs, security questionnaires — security@boxes.sh