Zero Data Retention
Zero Data Retention (ZDR) is a security commitment in which an AI system does not store, log, or reuse the data it processes after completing a task, so a request's inputs and outputs are not retained afterward.
When enterprise data flows through an AI model — for example, CRM records used by a Sales Assistant Agent to generate a summary or recommendation — Zero Data Retention means that data is used only to complete that specific task and is not saved, logged for training, or accessible afterward. This differs from standard Data Retention policies, which may keep logs or outputs for a defined period for auditing, debugging, or model improvement. ZDR is typically a requirement for enterprises handling sensitive customer or business data through third-party or embedded AI models, and is often paired with other AI Guardrails such as encryption in transit and access controls, alongside a broader security and trust program covering the option to run models in a private or on-premises environment.
Frequently Asked Questions
Zero Data Retention is a policy where an AI system does not store or reuse the data it processes once a task is complete.