Promptfoo

OpenAI acquired Promptfoo in March 2026 as part of a wave of 38 cybersecurity M&A transactions counted by SecurityWeek in that month alone. Promptfoo provided developer tools for testing large language model (LLM) applications against prompt-injection attacks, jailbreaks and other adversarial inputs, with an open-source product that had significant developer adoption.

Promptfoo was one of the most widely used open-source frameworks for adversarial testing of LLM applications, offering red-teaming capabilities alongside integration with popular AI development workflows. Its acquisition by OpenAI folds prompt-injection defence into the Frontier AI safety platform, giving OpenAI a tooling layer to verify that AI applications built on its models are resistant to adversarial manipulation.

The acquisition reflects a broader pattern: as AI applications enter production at scale, the attack surface of prompt injection and adversarial input has moved from theoretical research to an operational security concern. OpenAI's decision to acquire rather than build the tooling signals that Promptfoo had enough developer traction to be the reference implementation in the space.