Parlyreply has secured a Zero Data Retention (ZDR) agreement with OpenAI for its AI-powered features, adding an additional layer of protection for sensitive parliamentary casework data.
Under this configuration, data sent from Parlyreply to OpenAI is not stored, logged, or used to train OpenAI models. Prompts and responses are processed to generate an answer, and then discarded by the provider.
What the agreement covers
The ZDR setup applies to the AI features built into Parlyreply - including drafting assistance, case summaries, and smart suggestions. When these tools are used, casework content is streamed to OpenAI, a response is generated, and OpenAI does not retain that underlying data.
Within Parlyreply itself, only the information needed to power features and maintain a reliable audit trail is stored. This includes things like the final drafted text, metadata required for search, and logs necessary for accountability.
Why this matters for MPs’ offices
Constituency casework routinely includes medical information, immigration status, financial difficulties, and other highly sensitive details. For MPs and their staff, it is not acceptable for this material to be retained by third-party AI providers or repurposed to improve future models.
The ZDR agreement means offices can use Parlyreply’s AI tools while keeping casework out of training datasets and long-term logs at the model provider level.
Built for parliamentary standards, not generic SaaS
The ZDR setup sits alongside Parlyreply’s existing controls: role-based access, office-level scoping, and detailed audit trails for casework activity. The aim is to meet the expectations that already exist around parliamentary systems, rather than asking offices to relax them for AI.
Offices that want to review the details can request a walkthrough of how AI requests are handled end-to-end, including data flows, retention, and how ZDR is applied in practice.