OpenAI to acquire AI security platform Promptfoo to bolster enterprise agent safety
The AMW Read
Acquisition integrates a safety evaluation platform into a top-tier agent provider, updating the player map and exemplifying the acqui-licensing pattern, with segment-level impact on enterprise agent adoption.
OpenAI to acquire AI security platform Promptfoo to bolster enterprise agent safety
OpenAI has announced the acquisition of Promptfoo, an AI security startup specializing in penetration testing and safety evaluation for AI agents and LLM applications. The deal, whose terms were not disclosed, integrates Promptfoo's red-teaming and prompt-injection detection tools directly into OpenAI's enterprise agent stack, including the Agents SDK and the developer platform. Promptfoo co-founder Ian Webster will join OpenAI's safety team.
The acquisition accelerates OpenAI's enterprise security posture at a time when business adoption of autonomous agents is rising fast. By embedding Promptfoo's evaluation framework into its product, OpenAI gains a standardized safety testing layer that enterprise buyers increasingly demand before deploying agents in production. This move mirrors the "acqui-licensing" pattern seen elsewhere in the industry—acquiring a specialized tool to internalize a capability rather than building from scratch—and signals that agent safety is becoming a competitive differentiator for enterprise platform choice.
From a market perspective, the deal highlights how AI security tooling is shifting from standalone products into integrated features of major platform providers. Enterprises evaluating agent platforms must now weigh built-in safety guarantees against third-party add-ons. For OpenAI, the integration of Promptfoo into its developer toolchain could narrow the moat of competitors like Anthropic and Google, who also emphasize safety-by-design in their agent offerings. The pattern reinforces that safety, long a topic of academic debate, is now a tangible procurement requirement.
