OpenAI Moves to Bolster AI Security With Promptfoo Acquisition

OpenAI is moving deeper into the fast-growing field of AI security, announcing the acquisition of Promptfoo, a platform designed to help companies detect vulnerabilities in artificial intelligence systems before they reach production. The deal highlights a growing industry focus on securing AI models during development, as businesses rapidly deploy generative AI tools across critical applications.

According to OpenAI, Promptfoo’s tools are designed to identify and remediate weaknesses in AI during the development process, rather than after systems are deployed. The platform is aimed at enterprise users, with an emphasis on strengthening the security posture of AI applications before they reach production.

By bringing Promptfoo into its fold, OpenAI is signalling a specific interest in security practices that are integrated directly into AI development workflows. The focus on early detection and remediation of vulnerabilities reflects a growing recognition that AI models, like traditional software, can introduce risks if security is treated as an afterthought.

This acquisition positions Promptfoo’s capabilities alongside OpenAI’s broader AI offerings, with the stated goal of making it easier for enterprises to harden their AI systems against potential threats during the build phase.

Promptfoo has gained traction among developers and security teams for its ability to simulate attacks against AI models, helping organisations test how systems respond to adversarial prompts, prompt injections, data leakage attempts and other emerging threats unique to generative AI. By surfacing these vulnerabilities early, developers can refine prompts, adjust model behaviour, and implement guardrails before systems go live.

The move comes as businesses increasingly integrate AI into customer-facing products, internal automation tools, and decision-making systems. As adoption grows, so too have concerns around AI safety, data exposure, model manipulation, and misuse of automated agents.

Security experts have warned that large language models can be susceptible to prompt injection attacks, jailbreak attempts, and unintentional disclosure of sensitive information if they are not properly tested. Tools like Promptfoo attempt to address these risks by allowing teams to run automated security tests as part of their development pipelines.

For OpenAI, the acquisition also reflects a broader shift toward building a more complete enterprise AI stack, where development tools, safety testing, and deployment infrastructure work together within a single ecosystem.

While OpenAI did not disclose the financial terms of the deal, the company said Promptfoo’s technology will be integrated into its developer platform to help organisations build AI systems that are not only powerful but also secure and resilient from the outset.

The acquisition highlights the growing importance of AI security as a core discipline, as companies move from experimental AI pilots to full-scale production deployments across critical business operations.