Trust centre
Security at Danphe Council
What protects your documents today, what we have not done yet, and who processes your data.
Certifications
Where we are
No security certifications yet
- SOC 2 Type IIPlanned
- Independent penetration testPlanned
- ISO/IEC 27001Under consideration
Controls
What is in place today
Encryption per tenant
Data is encrypted in transit (TLS) and at rest. Stored files and sensitive fields use a separate encryption key for each organisation, managed in AWS KMS, so deleting an organisation's key makes its data unreadable.
Isolation between organisations
Every organisation's data is separated by PostgreSQL row-level security, forced on every tenant table. Automated tests try to read across organisations and must fail.
Tamper-evident audit chain
Each answer's audit record is hash-chained to the previous one and anchored with an asymmetric signing key. A verifier detects any later change to a record.
No training on your data
Your files, questions and answers are not used to train models, by us or by the model providers we use under their enterprise terms.
Upload checks
Uploaded files are quarantined and scanned for malware before they are read. Files that fail are rejected.
Roles and least privilege
Workspace roles limit who can change models, policies and settings.
Vendor controls
Workspace rules can restrict which model vendors may be used and require providers that do not retain prompts or outputs.
Secrets
Credentials are never stored in code. Model output is rendered without scripts or remote images.
Sub-processors
Who processes your data
| Company | Purpose | Data |
|---|---|---|
| Amazon Web Services | Hosting, storage, encryption keys; AI models through Amazon Bedrock | Account data, documents, questions and answers |
| Microsoft Azure | AI models through Azure OpenAI and Azure AI Foundry | Questions and the document passages needed to answer them |
The full list with locations is on the sub-processors page. We give 30 days' notice of changes to customers with a DPA.
Report an issue
Responsible disclosure
Found a vulnerability? Email trust@danphe.ai with the details and steps to reproduce. Please give us reasonable time to fix it before disclosure, and do not access other customers' data. We do not run a paid bug bounty yet.
Security questionnaires and the AI system card: see AI system card or contact us.