Security & data handling.
Ground Truth is an AI-native revenue workspace. We take the security and privacy of customer data seriously and design for it. This overview summarises our posture; our Data Processing Agreement (available on request) and sub-processor list provide the contractual detail.
Where your data lives
- Stored at rest in the EU (Stockholm, eu-north-1).
- Backend compute in the EU (Amsterdam, europe-west4).
- Some processing in transit — certain AI model calls, email delivery — occurs in the US or on global infrastructure under EU Standard Contractual Clauses (and the UK Addendum for UK customers). See the sub-processor list.
AI handling
- We use leading AI model providers — Google (Vertex AI), EU-hosted open-weights inference (TensorX, Dublin/Helsinki, zero data retention), and Anthropic — all contractually bound not to train on your data. Model evaluation runs on EU endpoints.
- We do not train our own models on customer data. If that ever changes, it will be opt-in only — with the customer's explicit written agreement, never by default.
- AI-drafted communications are sent only after a human reviews and approves them (supervised mode).
Encryption
- In transit: TLS for all connections.
- At rest: database encryption; OAuth tokens for connected mailboxes are additionally encrypted at the application layer, with the key required in production.
Access control & tenant isolation
- Row-level security with organisation-scoped policies on data tables.
- Every data access is scoped to the owning organisation (multi-tenant isolation), with role-based access (owner / manager / member) and least-privilege service credentials.
- Authentication via a managed identity provider; signed JWTs with issuer and audience pinning.
Application security
- Input validation and bounded pagination; sanitised error responses.
- SSRF protection on outbound requests; AI prompt-injection defences (external-data wrapping and anti-injection instructions); rate limiting.
- Internal security audit completed June 2026, with findings remediated.
Accountability & monitoring
- Record-change audit logging; AI-call tracing.
- Ingestion health monitoring with operator alerting.
- Breach notification to affected customers without undue delay and within 48 hours of becoming aware.
Government & public-authority requests
- Every public-authority request for customer data is reviewed for legal validity before any response; unlawful or overbroad requests are challenged.
- Any disclosure is limited to the minimum legally required, and each request — and our response — is documented.
- We notify the affected customer unless legally prohibited, and where possible redirect the authority to the customer directly.
- To date we have received no such requests.
Data subject rights & retention
- We assist customers with access, rectification, erasure, and export requests for identified data subjects.
- Customer data is retained for the duration of the subscription; on termination we delete or return it within 90 days.
Access to the platform
- Invite-only: there is no public self-serve signup. Accounts exist only for onboarded customers and their invited team members.
Connections you control
- We request least-privilege scopes for connected mailboxes (read + send) and CRM. You choose which mailboxes are connected, and you can disconnect at any time.
Security or privacy questions: [email protected]