OpenAI has announced its agreement to acquire Promptfoo Inc., a startup focused on testing and security evaluation for artificial intelligence applications. The terms of the acquisition were not disclosed.
Founded in 2024, Promptfoo evolved from an open-source framework to a commercial platform designed to evaluate AI prompts and model behavior. It is utilized by developers and enterprise security teams to test applications built on large language models and generative AI systems.
The platform addresses critical risks associated with deploying AI models, including prompt injection, data leakage, and jailbreak attacks. Its architecture enables systematic evaluation of AI system responses to structured inputs and adversarial prompts.
Promptfoo's technology allows teams to define prompts, expected outputs, and evaluation criteria in configuration files. The system then tests these prompts across various AI models and applications, scoring responses against predefined rules or automated grading.
Core functionalities include prompt testing, red-team simulations, and evaluation dashboards. The system also supports regression testing, enabling comparison of model outputs across versions and configuration changes.
OpenAI plans to integrate Promptfoo's technology into its OpenAI Frontier platform, which is used for building and operating AI coworkers. This integration aims to help enterprises test agent behavior, detect risks before deployment, and maintain oversight and accountability.
Srinivas Narayanan, CTO of B2B applications at OpenAI, stated that Promptfoo brings "deep engineering expertise in evaluating, securing and testing AI systems at enterprise scale." He added that their work "helps businesses deploy secure and reliable AI applications."