AI
[AD] Leaderboard 728×90 / 320×50
Back to feed
OpenAI5 min read

OpenAI Acquires Promptfoo — Security Is Now Built Into the Agent Platform

By AI Guide News·Monday, March 9, 2026
OpenAI Acquires Promptfoo — Security Is Now Built Into the Agent Platform

OpenAI is acquiring Promptfoo, an AI security startup trusted by 25% of Fortune 500 companies, to embed automated red-teaming and vulnerability testing directly into its Frontier enterprise platform.

[AD] Rectangle 300×250 / In-article

The Acquisition That Signals a Shift

On March 9, 2026, OpenAI announced it is acquiring Promptfoo — an AI security and evaluation startup founded by Ian Webster and Michael D'Angelo. The deal terms were not disclosed, but the signal is clear: as enterprises deploy AI agents into real workflows, security testing can no longer be an afterthought bolted on at the end. It needs to be built into the platform from the start.

Promptfoo's technology will be integrated directly into OpenAI Frontier, OpenAI's enterprise platform for building and operating AI coworkers, which launched just a month earlier in February 2026.

Why Promptfoo, Why Now

The numbers tell the story. Founded just two years ago, Promptfoo had already reached over 150,000 developers and was trusted by more than 25% of Fortune 500 companies. The company raised $23 million in total — including an $18.4 million Series A in July 2025 led by Insight Partners with Andreessen Horowitz participation — at an $86 million valuation. OpenAI moved to acquire it within eight months of that round closing.

This isn't OpenAI identifying a gap and building a solution. This is OpenAI acquiring the solution the market had already chosen. When a quarter of the Fortune 500 adopts a two-year-old startup's tooling, the adoption precedes the acquisition rationale.

What Promptfoo Does

Promptfoo specializes in automated security testing and evaluation for AI systems. Its platform allows developers to systematically simulate adversarial attacks directly within their development workflows, covering:

  • Prompt injection detection — catching attempts to hijack AI behavior through malicious inputs
  • Jailbreak identification — testing resistance to attempts to bypass safety guardrails
  • Data leak prevention — identifying where sensitive information might be exposed
  • Tool misuse detection — catching scenarios where agents use connected tools in unintended ways
  • Out-of-policy behavior monitoring — ensuring agents stay within defined boundaries

It also provides a widely used open-source CLI and library for testing LLM applications — a community tool that OpenAI has committed to continuing to maintain under the current license.

How It Fits Into Frontier

Frontier is OpenAI's enterprise platform for deploying "AI coworkers" — agents that operate within real business workflows with access to company data, tools, and systems. The more capable those agents become, the higher the stakes if they behave unexpectedly.

As Srinivas Narayanan, CTO of B2B Applications at OpenAI, put it: "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."

Once integrated, Frontier will offer security testing and red-teaming as a native part of the development workflow — not a separate step, but something that runs continuously as agents are built and modified. OpenAI also plans to add stronger oversight tools for tracking testing activity, monitoring changes over time, and maintaining compliance records.

The Open Source Commitment

One of the more notable aspects of this deal: OpenAI has explicitly committed to keeping Promptfoo's open-source project alive and continuing to serve its existing customers. In a market where acquisitions often mean the end of an open-source tool's community life, that commitment matters — and it reflects the reality that the developer community's trust in Promptfoo was part of what made the acquisition valuable in the first place.

The Pattern Emerging

Promptfoo is the latest in a series of targeted acquisitions: healthcare tech startup Torch earlier this year, AI interface maker Software Applications before that, and developer tool creator Peter Steinberger in January. The pattern is consistent — OpenAI is filling enterprise capability gaps through acquisition rather than waiting to build. In a market where enterprise AI adoption is accelerating faster than any single team can build, that's a rational strategy. The question is whether these pieces, once integrated, form a coherent platform — or a collection of parts.

Source: https://openai.com/vi-VN/index/openai-to-acquire-promptfoo/

openaipromptfoosecurityfrontierai-agentsacquisitionenterprise
[AD] Leaderboard 728×90 / end of article