OpenAI announced that Zero Data Retention (ZDR) is now available for its frontier models through the API, for eligible customers. Under ZDR, OpenAI does not retain the content of API requests and responses after they are processed — a stricter posture than OpenAI's standard retention policy, which by default stores API data for a limited period (typically 30 days) for abuse monitoring unless customers opt out.
ZDR itself is not new — OpenAI has offered it for some products previously — but this expansion applies the policy to its most capable, frontier-tier models via the API, closing a gap that had pushed some enterprise customers toward older or lower-tier models when handling sensitive data.
For B2B companies building sales, support, or operations automation on top of OpenAI's API, data retention has been a recurring compliance sticking point, particularly for firms in finance, healthcare-adjacent services, or those with EU customers subject to GDPR data minimization requirements. A no-retention option gives compliance and security teams a cleaner story: prompts containing customer names, deal terms, or support ticket contents are processed and then not stored by the model provider.
Eligibility details are limited in the announcement — it's unconfirmed which specific frontier models, API tiers, or account types qualify, and whether ZDR applies uniformly across features like function calling, file search, or the Assistants API. Companies interested in ZDR should confirm scope and eligibility directly with OpenAI or their account representative rather than assuming coverage.
The move follows a broader industry pattern of AI vendors adding enterprise-grade data controls — retention policies, regional data residency, and audit logging — as a condition for landing larger, more risk-averse customers. For smaller B2B operators, the practical effect is a lower barrier to using frontier-tier models for workflows that were previously restricted to on-premise or heavily redacted data.