OpenAI announced it is acquiring Promptfoo, a platform that scans AI models for security flaws during development. The deal was disclosed on March 9, 2026. Promptfoo lets enterprise teams run automated tests that surface issues such as prompt injection, data leakage, and model‑steering risks. OpenAI plans to integrate the tool into its own developer offerings.
The move signals OpenAI’s acknowledgement that AI security is becoming a bottleneck for adoption. As more companies embed large language models into products, the attack surface widens. By bundling a dedicated testing suite, OpenAI hopes to lower the barrier for safe deployment and differentiate its stack from rivals that still rely on third‑party auditors. It also gives the firm a foothold in the emerging AI‑risk tooling market, which has seen recent funding rounds for startups like Snyk AI and Guardrails AI.
OpenAI’s purchase mirrors a broader trend of AI leaders buying security specialists rather than building in‑house solutions. Microsoft’s 2025 acquisition of a model‑hardening startup, for example, aimed at the same pain point. Whether Promptfoo’s approach scales to OpenAI’s rapid model iteration remains to be seen, but the integration will likely push developers toward more structured testing before release.
In short, OpenAI is not just adding a product; it is signaling that AI safety will be a separate line item on every development budget.