OpenAI Reaffirms Zero Data Retention, Previews Private Safety Processing
OpenAI has reaffirmed its Zero Data Retention policy for eligible API customers, ensuring that data from frontier models is not retained. The company also previewed Private Safety Processing, a mechanism designed to enable advanced AI safety testing without compromising data privacy. This announcement underscores OpenAI's commitment to data security while advancing safety protocols for its most powerful models. The new processing approach aims to reconcile safety analysis with strict data-handling guarantees. Specific implementation details and timelines were not disclosed.
Key facts
- OpenAI reaffirms Zero Data Retention for eligible API customers.
- The policy applies to frontier models.
- Private Safety Processing was previewed by OpenAI.
- Private Safety Processing enables advanced AI safety analysis.
- The approach does not compromise data privacy.
- OpenAI aims to balance safety testing with strict data-handling guarantees.
- The announcement emphasizes OpenAI's commitment to data security.
- No implementation details or timelines were provided.
Entities
Institutions
- OpenAI