AI Signal 470
OpenAI launches Private Safety Processing for zero data retention, contrasting Anthropic's 30-day retention for top models
↗
OpenAI announced Private Safety Processing, an automated mechanism that scans customer interactions for safety risks without exposing prompts or responses to personnel, enabling zero data retention where Anthropic still requires 30-day retention for its top models.
Enterprises evaluating AI providers now have a meaningful privacy distinction between the two leading frontier model companies. Anthropic retains prompts and outputs for 30 days on its top models for safety review, while OpenAI claims its automated system can perform safety checks without personnel ever accessing the content.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
Private Safety Processing scans for misuse patterns across related interactions without exposing prompts or responses to OpenAI personnel
Anthropic requires 30-day data retention for its covered models Mythos 5 and Fable 5 as part of its ZDR offering
OpenAI limits human review to child exploitation material detection, while Anthropic allows human review of any content flagged by automated systems
THE CLUSTER