Grouped story
OpenAI Unveils Private Safety Processing System
OpenAI has launched a new privacy-centric safety system called Private Safety Processing, designed to prevent misuse of its AI technology without retaining customer data. This innovative system identifies patterns of abuse while ensuring that customer data remains encrypted and under the control of the users, contrasting with competitor Anthropic, which retains user data for 30 days for safety checks.
Key points
Private Safety Processing analyzes multiple conversations for signs of misuse without human review.
A technical white paper detailing the new safety system is expected in September 2026.