Reuters reports that most of OpenAI's code resides in a single monorepo, with employees warning about security vulnerabilities for years.

According to Reuters, OpenAI stores most of its code in a single monorepo. Any employee with code access can see details on many projects, including sensitive ones unrelated to their own work. Employees have warned senior leadership about this issue for years, with complaints reaching executives and the board, though some staff believe keeping pace with Google and Anthropic is more urgent. OpenAI states that model weights and most sensitive work are stored outside the monorepo.