Security

GovRedact handles sensitive personal data for UK public authorities, so security is built into the platform. This statement summarises the technical and organisational measures we maintain. It forms part of our Data Processing Agreement.

1. Certifications

CloudyIT holds Cyber Essentials Plus certification. Our hosting provider, Microsoft Azure, maintains ISO 27001, SOC 2 and UK government-recognised assurance for its data centres.

2. Hosting and data residency

GovRedact runs on Microsoft Azure in the UK South and UK West regions. Customer Data is stored in the United Kingdom. Infrastructure is provisioned as code and runs on Azure Container Apps with managed identities rather than long-lived secrets where possible.

3. Encryption

• Data in transit is encrypted with TLS 1.2 or higher.
• Data at rest is encrypted using Azure platform-managed keys.
• Application secrets are stored in Azure Key Vault and accessed via managed identity.

4. Access control and authentication

• Sign-in is via Microsoft Entra ID single sign-on; we encourage MFA in your tenant.
• Role-based access control limits what each user can do, with separation-of-duties checks on sensitive actions (such as maker–checker on disclosure).
• Sensitive administrative actions require step-up re-authentication.
• Each organisation's data is logically isolated and bound to its own Entra tenant.
• CloudyIT staff do not have standing access to Customer Data; elevated access is just-in-time and audited.

5. Malware protection

Uploaded files are scanned for malware before processing. The platform can be configured to reject uploads if the scanner is unavailable (fail-closed).

6. Logging and audit

Security-relevant events are recorded in an immutable, hash-chained audit log. Operational telemetry is collected via Azure Application Insights for reliability and security monitoring.

7. Resilience and backups

Managed Azure services provide redundancy, and our database uses automated backups with geo-redundancy in production. We test our ability to restore service.

8. Secure development

Changes go through code review and automated checks before deployment. Dependencies are kept up to date and infrastructure is defined and reviewed as code.

9. AI processing

AI-assisted redaction uses Anthropic's Claude models. Content sent for processing is used only to return a result and is not used to train third-party models. AI output is always subject to human review before disclosure.

10. Reporting a vulnerability

If you believe you have found a security issue, please contact security@cloudyit.co.uk. We welcome responsible disclosure and will work with you to investigate and resolve genuine issues.