OpenAI Acquires Promptfoo to Bolster AI Agent Security in Frontier Platform

March 9, 2026

Details

  • OpenAI announced the acquisition of Promptfoo, an AI security startup founded in 2024 by Ian Webster and Michael D’Angelo, to enhance security testing for AI agents.
  • Promptfoo's technology will integrate into OpenAI Frontier, the enterprise platform for building and operating AI coworkers launched in early February 2026.
  • The tools enable automated red-teaming, evaluating agentic workflows for risks like prompt injections, jailbreaks, data leaks, tool misuse, and out-of-policy behaviors.
  • Promptfoo serves over 25% of Fortune 500 companies, with more than 125,000 developers using its open-source CLI and library; OpenAI commits to keeping it open source and supporting existing customers.
  • Total funding for Promptfoo was about $23 million, including a $5 million seed from Andreessen Horowitz and an $18.4 million Series A in July 2025, valuing it at $86 million.
  • New Frontier capabilities include security testing built into the platform, workflow integration for risk remediation, and reporting for governance and compliance.

Impact

OpenAI's acquisition of Promptfoo positions it to address a critical vulnerability in the booming AI agent market, where autonomous systems accessing enterprise data and tools amplify risks of exploits like jailbreaks and data leaks. This move integrates enterprise-grade red-teaming directly into Frontier, launched just weeks ago, helping clients like Uber and Intuit deploy agents safely amid rising regulatory scrutiny on AI governance. It pressures rivals such as Anthropic and Google DeepMind, which have emphasized safety but lack comparable native agent security testing in their platforms; Promptfoo's adoption by over 25% of Fortune 500 firms gives OpenAI an edge in enterprise trust. By maintaining the open-source project, OpenAI fosters developer ecosystem buy-in while advancing closed capabilities, potentially accelerating agent adoption curves and steering R&D toward secure, compliant AI coworkers over the next 12-24 months. This aligns with trends in AI safety benchmarks and on-device inference security, narrowing gaps in production-ready agent reliability.

Read more