OpenAI buys Promptfoo to bolster Frontier AI security

OpenAI has agreed to acquire Promptfoo, an AI security platform that tests and evaluates AI systems during development. The company plans to integrate Promptfoo's technology into OpenAI Frontier, its platform for building and operating AI coworkers.

The deal adds specialised tools for enterprises deploying agent-based systems into day-to-day workflows. As more companies roll out these deployments, attention has intensified around evaluation, security, and compliance controls for AI behaviour-especially when systems interact with business data, internal tools, and external services.

Promptfoo is known for tools that test large language model applications and support red-teaming exercises. Its products include an open-source command-line interface and library that developers use to evaluate systems and probe for weaknesses.

Frontier focus

Frontier sits at the centre of OpenAI's enterprise push around AI coworkers, which it positions as agents that carry out tasks across business processes. Promptfoo's technology is expected to become a native part of Frontier once the acquisition closes.

As companies move beyond pilots and connect AI systems to live workflows, they must manage the risk of errors and unwanted actions. They also face governance requirements for how systems are tested, how changes are made over time, and how decisions can be reviewed.

OpenAI described evaluation, security, and compliance as foundational requirements for enterprise deployments, citing the need for systematic testing of agent behaviour, risk detection before deployment, and record-keeping for oversight and accountability.

Testing tools

Promptfoo's toolset is designed to surface vulnerabilities that can appear in AI applications built on large language models. OpenAI said the combined offering will target issues such as prompt injections, jailbreaks, data leaks, tool misuse, and out-of-policy agent behaviour.

The integration is expected to bring automated security testing and red-teaming functions into Frontier. This work typically covers adversarial inputs, attempts to override system instructions, and probes for unintended disclosure of sensitive information.

OpenAI also plans to integrate security and evaluation more directly into development workflows. It said earlier identification, investigation, and remediation of agent risks will be a core part of how enterprise AI systems are developed and operated.

Another element is reporting and traceability. Integrated reporting is expected to document testing and track changes over time, helping organisations meet governance, risk, and compliance expectations for AI.

Open source plan

Promptfoo's team is led by Ian Webster and Michael D'Angelo. OpenAI said Promptfoo's tools are used by more than 25% of Fortune 500 companies.

OpenAI plans to continue developing Promptfoo's open-source project alongside integrated enterprise features inside Frontier. Open-source tools are a common way for security and testing products to reach developer teams, and can later expand into paid offerings with broader management and reporting features.

"Promptfoo brings deep engineering expertise in evaluating, securing, and testing AI systems at enterprise scale. Their work helps businesses deploy secure and reliable AI applications, and we're excited to bring these capabilities directly into Frontier," said Srinivas Narayanan, CTO of B2B Applications at OpenAI.

Promptfoo was founded to address the need for practical ways to secure AI systems as developers began putting language models into production. The challenge has grown as AI agents gain access to proprietary data, software tools, and operational systems.

"We started Promptfoo because developers needed a practical way to secure AI systems. As AI agents become more connected to real data and systems, securing and validating them is more challenging and important than ever. Joining OpenAI lets us accelerate this work, bringing stronger security, safety, and governance capabilities to the teams building real-world AI systems," said Ian Webster, Promptfoo's co-founder and CEO.

Completion of the transaction remains subject to customary closing conditions.