OpenAI previewed Private Safety Processing, a new system that detects AI safety risks across multiple interactions without retaining customer content or violating Zero Data Retention guarantees.

Key Details

• Private Safety Processing evaluates patterns across related interactions without giving OpenAI personnel access to customer content
• Supports customer-controlled infrastructure as well as OpenAI storage encrypted with customer-managed keys
• Full rollout and technical white paper planned for September 2026

In This Story