OpenAI acquires Promptfoo to strengthen security for AI agents

OpenAI said on Monday that it has acquired Promptfoo, an AI security startup launched in 2024 to help protect large language models from online adversaries.

In a blog post, the Frontier AI lab said that after the transaction closes, Promptfoo’s technology will be folded into OpenAI Frontier, the company’s enterprise platform built for AI agents.

The rise of increasingly autonomous AI agents capable of performing digital tasks has fueled excitement about potential productivity gains. At the same time, it has opened up new opportunities for malicious actors to gain access to sensitive information or interfere with automated systems. The acquisition highlights how frontier AI companies are moving quickly to show that their technology can be deployed safely in high-stakes business environments.

Pian Webster and Michael D’Angelo founded Promptfooto to create tools that allow companies to test large language models for security vulnerabilities, including through an open-source interface and library. The startup says its products are used by more than 25% of Fortune 500 companies.

Since its founding, Promptfoo has raised only $23 million and was valued at $86 million following its latest funding round in July 2025, according to PitchBook. OpenAI did not reveal the financial terms of the acquisition.

OpenAI stated in its post that Promptfoo’s technology will enable its agent platform to perform automated red-teaming, assess agentic workflows for security risks, and monitor activity for compliance and other risk-related requirements. The company also said it expects to keep expanding Promptfoo’s open-source offering.