GenAI security statement

We prioritise the security of our clients’ information

Write Group’s AI+Human™ services include training and content work. We prioritise the security and privacy of our clients’ information, including the content clients entrust to us. As we continue to integrate AI tools into our services, we are committed to maintaining comprehensive security controls.

We use a bespoke platform with multiple large language models

Our AI+Human™ services are delivered through the LibreChat platform, which provides secure access to multiple generative AI tools, including Azure OpenAI GPT-4, OpenAI API, Google Gemini, Anthropic Claude, and other leading large language model (LLM) providers.

Our platform uses enterprise-grade security safeguards

The LibreChat platform implements comprehensive security controls, including multifactor authentication through Microsoft Entra ID, AES-256 data encryption at rest, TLS 1.2/1.3 encryption in transit, complete user data isolation, and role-based access controls. The platform is hosted on ISO 27001- and SOC 2-certified infrastructure with automated backups and disaster recovery capabilities.

Our platform does not train LLMs

We take the security and privacy of client documents very seriously. The LibreChat platform does not train any LLMs and never will. All application programming interface (API) keys are encrypted and stored securely, with complete separation between user accounts, ensuring no data mixing between clients or organisations.

Our platform provides flexible data retention controls

The platform is configured to work with AI providers that offer zero-data-retention options, where available. However, similar to other cloud-based AI services, some LLM providers may retain data securely for a defined period of time for operational reasons, such as:

• legal or regulatory requirements — including monitoring for misuse, ensuring adherence to terms of service, and maintaining audit trails for compliance purposes
• abuse monitoring and policy enforcement — retaining short-term logs of inputs and outputs to enable detection and investigation of spam, malware, policy violations, and inappropriate content
• performance metrics and quality control — using aggregated and anonymised usage data to identify system issues, latency problems, or service performance regressions
• session continuity and security — temporarily storing session data to allow for secure user authentication, conversation history, and audit logging, while maintaining platform security and functionality.

For Copilot users — Azure OpenAI API for Microsoft security on request

For organisations using Microsoft Copilot Enterprise, we have configured our platform to access AI services exclusively through Azure OpenAI Service APIs. This option is available on request.

When using our Azure OpenAI Service, your data benefits from Microsoft’s enterprise-grade security infrastructure, zero-day retention policy, and compliance certifications such as HIPAA, FedRAMP, and SOC 2.

This means that, while LibreChat provides the user interface and platform functionality, all AI processing occurs within Microsoft’s Azure infrastructure — the same trusted environment that powers other Microsoft Enterprise services.

If you need this service, please let us know in writing.