OpenAI to acquire Promptfoo and keep its AI security stack open source

On March 9, 2026, OpenAI said in an X post that it plans to acquire Promptfoo, an AI security platform used to evaluate and red-team LLM applications. The company framed the move as an expansion of OpenAI Frontier, its enterprise stack for building and deploying safer agents.

OpenAI’s announcement says Promptfoo is already used by more than 25% of Fortune 500 companies to assess AI application security. The company said the acquisition is meant to add integrated security and safety testing, tighter development workflow integration, and stronger oversight and accountability features for teams shipping agentic systems at scale.

• Promptfoo’s open-source project will remain under the current license.
• OpenAI said current customers will continue to be serviced and supported.
• The transaction is still subject to customary closing conditions.

That matters because the center of gravity in AI product work has shifted from simple chat interfaces to agents that browse the web, call tools, write code, and touch production systems. In that environment, evaluation is not only about answer quality. It also includes prompt injection resistance, tool misuse, data leakage, policy enforcement, and reproducible red-teaming before and after deployment.

The immediate signal is that OpenAI wants application security testing to sit much closer to model deployment and agent operations rather than remain a separate afterthought. The original X post is here, and the company announcement is on OpenAI.