OpenAI Research Chief Says It Uses De-Identified Metadata Amid Data Privacy Concerns

Concerns over whether AI companies can learn from enterprise customer data are reshaping the AI business landscape. On September 14, OpenAI Chief Research Officer Mark Chen responded on X to data usage concerns, stating that the company uses de-identified metadata that does not include information capable of identifying customers.

OpenAI has recently been accused of potentially using user data to train models in pursuit of major mathematical breakthroughs. In parallel, after Anthropic adjusted the data retention policy for its flagship model Fable in June, major enterprises including Nvidia, Palantir, and Booz Allen Hamilton have restricted or reduced their use of Fable and demanded irreversible zero-data-retention guarantees from Anthropic. Microsoft has meanwhile stepped in, pitching its own AI products and isolated cloud environments to security-conscious enterprises.