OpenAI previews Private Safety Processing for Zero Data Retention

OpenAI previewed Private Safety Processing, a safety approach designed to identify risk patterns across related interactions while remaining compatible with Zero Data Retention deployments.

Key points:

• OpenAI says eligible API customers using ZDR do not have prompts or responses retained after request processing.
• Private Safety Processing is intended to detect multi-interaction misuse without giving OpenAI personnel access to the underlying content.
• The company says customer-controlled infrastructure or customer-controlled encryption keys can preserve privacy while automated systems return limited safety signals.

Source: https://openai.com/index/offering-zero-data-retention-for-frontier-models/