Security
Openly is designed for private, sensitive conversations.
Chaperone uses layered technical and organizational controls to protect information, restrict access, monitor for misuse, and support recovery when something goes wrong. This page intentionally avoids certification or audit-status claims that have not been independently established.
No Third-Party Advertising or Session Replay
Openly does not use third-party advertising, cross-site tracking, or session-replay technologies in the user-facing product. Product analytics are first-party, and the event schema cannot carry conversation text or user-entered support content.
Encryption
Data in transit is protected with industry-standard transport encryption across production connections. Data at rest is encrypted across managed database and file-storage infrastructure. Application secrets are stored in environment-scoped secret storage and production credentials are separated from non-production.
Least-Privilege Access
Administrative and privileged access requires strong authentication controls. Chaperone uses granular role-based access controls and time-bounded elevated access so authorized users receive only capabilities appropriate to their role and operational need.
Audit Trail
Sensitive actions and permission events are recorded in an append-only audit trail with actor, role, affected record, and timestamp information. The trail is append-only at the database level and each entry is cryptographically hashed for tamper evidence.
Logging and Monitoring
Platform, application, and database activity is logged by our hosting and database providers and reviewed during incident response. We do not send application errors, conversation text, or other support content to third-party error-monitoring or session-replay services.
Production Isolation, Backups, and Recovery
Production is separated from non-production environments with separate credentials and secrets. Backups are encrypted and our managed database supports point-in-time restore.
AI Security and Governance
AI features are delivered through a governed provider path. Every request to our primary AI provider carries a provider-level instruction excluding providers that log or train on prompts, and that exclusion is applied on the request path by default rather than left to individual developer choice. Safety classification and output validation operate separately from individual Companion style.
Openly does not use member conversations for internal model improvement unless the user has provided the separate, affirmative permission for that use. When permission is active, eligible content must pass through the approved de-identification and minimization process before it is released into an internal improvement dataset.
Vendor and Service-Provider Controls
Service providers are evaluated based on the information they process and the function they perform. Providers handling sensitive information are subject to contractual privacy, confidentiality, security, retention, and deletion requirements as those agreements are completed, and integrations are designed to receive only information reasonably necessary for their function.
Vulnerability Management and Incident Response
Chaperone maintains an information-security incident-response process for the shared platform, and automated build and test pipelines include access-control and row-level-security testing. Public statements about penetration testing, recovery objectives, or other specific controls are made only when technically verified for the Openly production deployment.
Report a security concern or privacy question to privacy@chaperone.health.