OpenAI to Acquire AI Security Startup Promptfoo to Bolster Frontier Platform

OpenAI plans to incorporate Promptfoo's security and evaluation capabilities directly into its Frontier platform. This move is intended to provide enterprises with systematic tools to test AI agent behavior and identify potential risks before they are deployed into professional workflows. By embedding these features, OpenAI aims to address foundational requirements for security, compliance, and oversight in the operation of AI coworkers.

Promptfoo, led by Ian Webster and Michael D'Angelo, currently provides a suite of tools utilized by more than a quarter of Fortune 500 companies. The startup is also known for its widely adopted open-source library and command-line interface used for red-teaming applications powered by large language models. OpenAI has stated that it will continue to support the open-source project while simultaneously advancing integrated enterprise features within the Frontier environment.

The integration will introduce native security testing to Frontier, helping businesses detect and address vulnerabilities such as prompt injections, jailbreaks, data leaks, and unauthorized tool usage. Additionally, the platform will offer enhanced reporting and traceability to assist organizations in documenting their testing processes and meeting evolving governance and regulatory expectations. Srinivas Narayanan, OpenAI's CTO of B2B Applications, noted that the acquisition brings specialized engineering expertise in securing AI systems at an enterprise scale.

Promptfoo CEO Ian Webster emphasized that joining OpenAI will accelerate the development of safety and governance tools as AI agents become more deeply connected to sensitive data and systems. The finalization of the acquisition is subject to standard closing requirements.