Offering Zero Data Retention for frontier models

OpenAI has reaffirmed its commitment to Zero Data Retention for eligible API customers, indicating that data submitted through these APIs will not be stored for model training or other purposes. Additionally, the organization has previewed 'Private Safety Processing,' a new capability designed to enhance AI safety without compromising user data privacy.

Why it matters

This development addresses critical concerns around data privacy and the use of proprietary information when interacting with advanced AI models. It can foster greater trust and adoption among organizations hesitant to integrate AI due to data governance risks, potentially broadening the applicability of frontier models across sensitive sectors.

Key insights

• OpenAI is maintaining a policy of Zero Data Retention for qualifying API customers.
• Data submitted via eligible APIs will not be utilized for model training.
• A new feature, Private Safety Processing, is being introduced to advance AI safety protocols.
• Private Safety Processing aims to enhance safety measures while upholding data privacy standards.

Source

OpenAI Research — https://openai.com/index/our-commitment-to-zero-data-retention