Private Safety Processing means safety checks themselves don't see data — major play for healthcare & finance compliance.
OpenAI Reaffirms Zero Data Retention for API Customers, Previews Private Safety Processing
Original: Offering Zero Data Retention for frontier models
Importance: 多くのAPI利用企業のデータプライバシー懸念を直接解決し、コンプライアンス要件が厳しい業界での選定に大きく影響する。
Summary
OpenAI reaffirms Zero Data Retention for eligible API customers and previews Private Safety Processing, enabling advanced AI safety checks without storing or compromising user data privacy. This allows enterprises and developers to use frontier models with sensitive data while meeting regulatory compliance requirements.
Key Points
- Zero Data Retention reaffirmed for API-eligible customers; prevents both training data inclusion and server-side logging
- Private Safety Processing ensures safety filtering itself does not expose sensitive data
- Compliance barriers in regulated sectors (healthcare, finance, legal) significantly reduced
- Preview stage; GA timeline and implementation details remain unspecified
- Minimal expected impact on existing Zero Data Retention users
View developer notes (APIs, breaking changes, migration)
Zero Data Retention applies to API-eligible models, preventing request/response data from being added to training datasets or stored in server logs. Private Safety Processing preview suggests safety filtering will run in encrypted or secure enclave environments without exposing sensitive content. Specific parameter names, response formats, timeline, and affected model roster not disclosed; gradual GA rollout expected. Existing Zero Data Retention customers may require no additional configuration.
Source: https://openai.com/index/offering-zero-data-retention-for-frontier-models
Outlet: OpenAI News
This article is an AI-generated summary (OpenAI GPT-4o-mini) of publicly available information from Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek, Sakana, and other vendors. The original source URL is always provided in accordance with fair-use citation requirements. Summaries are AI-generated and may contain mistranslations or misinterpretations. Always verify details with the original source.