OpenAI acquires Promptfoo to secure its AI agents

OpenAI introduced Monday it has acquired Promptfoo, an AI safety startup based in 2024 to guard LLMs from on-line adversaries.

The frontier lab stated in a weblog submit that when the deal closes, Promptfoo’s know-how will probably be built-in into OpenAI Frontier, its enterprise platform for AI brokers.

The event of impartial AI brokers that carry out digital duties has generated pleasure about productiveness features. However it’s additionally given unhealthy actors contemporary alternatives to entry delicate information or manipulate automated methods. This deal underscores how frontier labs are scrambling to show their know-how can be utilized safely in important enterprise operations.

Promptfoo was based by Ian Webster and Michael D’Angelo to develop instruments that firms can use to check safety vulnerabilities in LLMs, together with an open supply interface and library. The corporate stories that its merchandise are utilized by greater than 25% of Fortune 500 firms.

Promptfoo has raised simply $23 million since its founding and was valued at $86 million after its most up-to-date spherical in July 2025, in accordance with PitchBook. OpenAI didn’t disclose the worth of the transaction.

OpenAI’s submit stated Promptfoo’s know-how will enable its agent platform to carry out automated red-teaming, consider agentic workflows for safety issues, and monitor actions for dangers and compliance wants. The corporate additionally stated it expects to proceed constructing out Promptfoo’s open supply providing.