OpenAI previews Private Safety Processing to keep zero retention

OpenAI said on August 19 it will keep offering Zero Data Retention for frontier models to eligible API customers and previewed Private Safety Processing, a system meant to spot misuse patterns across related interactions without giving OpenAI staff access to the underlying prompts or responses. Customer content would be encrypted with keys OpenAI does not hold, with automated systems producing only narrow safety signals; the company plans to start rolling the system out in September 2026 alongside a technical white paper. The move contrasts with rival Anthropic's mandatory 30-day retention policy for its most capable models, positioning zero retention as a selling point for enterprise customers.