Security & trust
Security and trust.
Security at Jiber means each AI process agent is scoped to the minimum data it needs, runs with least-privilege access, logs every action, and requires human approval before irreversible steps.
How Jiber designs agents to handle business data carefully. These are our working commitments; formal certifications are noted as planned where relevant.
Principles
Our commitments.
- Data minimisation. Each agent is scoped to the systems and fields its process needs, and nothing more.
- Your data, your purpose. Customer data is used only to run that customer’s agent, not to train shared models.
- Hosting by agreement. Deployment location and hosting provider are agreed per engagement, including options to run inside the customer’s own cloud account.
- Least-privilege access. Service accounts and staff access follow least privilege, with credentials stored in a secrets manager and rotated.
- Audit logs. Every agent action and decision is logged with inputs, rule or model used, and outcome, so it can be reviewed.
- Human approval. Customer-defined checkpoints require a person to approve outgoing messages, filings or other irreversible steps.
- Encryption. Data is encrypted in transit and at rest using standard cloud provider controls.
Certifications
Compliance roadmap.
Jiber does not currently hold formal security certifications. The following are planned as the company grows:
- SOC 2 Type II — planned.
- ISO/IEC 27001 — planned.
- India DPDP Act, 2023 — processes designed with its requirements in mind; reviewed with counsel per engagement.
Contact
Questions.
Security questions or a vendor questionnaire? Email hello@jiber.ai.