OpenAI is beefing up its AI defenses in its own backyard, announcing today that it is acquiring Promptfoo, a San Mateo startup whose open source toolkit helps developers red-team and secure AI agents. The Promptfoo engineering team and toolkit are headed to OpenAI’s Frontier group, while the company says the open source project itself will stay available to the wider community. Financial terms of the deal were not disclosed.
Integration Into Frontier
OpenAI said it plans to plug Promptfoo’s technology straight into OpenAI Frontier, the enterprise platform it offers for building, deploying and managing AI agents. In a blog post, OpenAI described the move as a way to strengthen agentic security testing and evaluation capabilities inside Frontier, positioning Promptfoo’s tools as a native part of how customers test and harden their agents.
What Promptfoo Does
Promptfoo’s core product is an open source framework that automates red-teaming, prompt evaluation and workflow tracing for AI agents. Teams can wire it into their CI/CD pipelines so they can systematically probe for prompt injection, data leakage and misuse of external tools before those issues make it into production. Documentation from Promptfoo highlights features for tracing an agent’s behavior across tool calls and sharing those traces for review, so security and compliance teams can replay and inspect what actually happened.
The startup has also talked up broad adoption among developers and said it crossed the 100,000-user mark, a milestone covered by PR Newswire.
Why The Buy Makes Sense
As AI agents get permission to call external tools and run multi-step workflows, the security problem shifts from testing individual prompts to evaluating entire end-to-end processes. Bloomberg Law reported that Promptfoo’s tooling can automate much of that red-teaming and highlight risky behaviors that might matter for compliance reviews. Recent academic work has mapped out new attack patterns against these agentic systems, underscoring the need for systematic testing, as shown in an arXiv preprint.
Funding And Ownership
Promptfoo raised $18.4 million in a Series A round in July 2025 led by Insight Partners with participation from Andreessen Horowitz, according to the company’s own announcement. Promptfoo said at the time that the money would go toward expanding both its product and its open source work.
PitchBook data cited in other coverage show slightly different totals for Promptfoo’s funding and valuation, a reminder that private-market numbers are often more fuzzy than precise. TechCrunch reported those PitchBook figures in its write-up of the deal.
Local Impact
Promptfoo is based in San Mateo and was co-founded by Ian Webster and Michael D’Angelo, according to reporting from CRN. Folding a local security-focused shop into the Frontier team deepens OpenAI’s Bay Area talent bench as the company continues to expand its footprint in the region. OpenAI’s Mission Bay expansion was previously detailed by a local outlet, which has tracked the company’s growing presence around San Francisco.
OpenAI has framed the Promptfoo acquisition as part of a broader push to make AI agent deployments more auditable and testable for enterprise customers, while keeping the Promptfoo project open and supporting multiple model and provider choices. In its announcement, OpenAI said the Promptfoo team will continue building inside the Frontier organization once the deal closes.
