OpenAI is testing a new safety system called Private Safety Processing that detects misuse patterns across multiple AI interactions without exposing customer data. The company began trials with early customers and plans a rollout in September, according to medianama.com. This system builds on OpenAI’s Zero Data Retention option, which allows eligible API users to exclude prompts and responses from data retention.
Private Safety Processing works by automatically identifying misuse patterns that may only be visible when analyzing multiple requests collectively. It achieves this without giving OpenAI employees access to the underlying customer prompts and responses. The system is designed to enhance safety while preserving user privacy, addressing concerns about data control and security in AI deployments.
The launch comes amid regulatory scrutiny in India, where the Digital Personal Data Protection Act, 2023, assigns responsibility to Data Fiduciaries for processing carried out by Data Processors. The Act’s relevant provisions will take effect on May 13, 2027. Additionally, the Reserve Bank of India is considering data governance rules for regulated entities, including third-party arrangements, though these remain in draft form. OpenAI’s approach aligns with these emerging data protection frameworks by emphasizing contractual safeguards and data privacy.
OpenAI’s Private Safety Processing system is expected to begin wider deployment in September, marking a significant step in balancing AI safety with enterprise data privacy requirements. The RBI’s draft guidance on data governance continues to be reviewed, with no binding regulations issued yet, highlighting ongoing regulatory developments in India’s AI data landscape.