OpenAI is acquiring Promptfoo, a tool enterprises use to find security holes in AI systems before they ship.
Promptfoo built its reputation as an open-source evaluation framework that lets developers probe AI applications for prompt injection, data leakage, and related vulnerabilities at build time - catching problems before they reach production rather than after. The platform is aimed squarely at enterprise teams trying to harden AI-powered products before deployment. Financial terms were not disclosed.
Every major AI lab is competing for enterprise contracts, and security has become a credible differentiator - no procurement team will route sensitive business data through a model it cannot demonstrate is safe. Buying Promptfoo lets OpenAI pitch a more complete stack: not just the model, but the testing infrastructure to prove it is fit for production.
The irony is hard to miss: OpenAI now owns a tool built to surface vulnerabilities in AI systems, including, one assumes, its own.