OpenAI to buy cybersecurity startup Promptfoo to help enterprises detect vulnerabilities in AI systems
Publish Date: 2026-03-09 23:57:00
Source Domain: www.storyboard18.com
Using an unordered list, summarize the following article with between 4 and 8 key points. OpenAI has agreed to acquire Promptfoo, a cybersecurity startup that develops tools to test and secure artificial intelligence systems during development, as the company expands safeguards for enterprise AI deployments.The ChatGPT maker said Promptfoo’s technology will be integrated into OpenAI Frontier, a recently launched platform designed to help organisations build and manage AI agents. OpenAI refers to these agents as “AI coworkers,” which can perform tasks across enterprise workflows with minimal human intervention. Financial terms of the deal were not disclosed, and the transaction remains subject to customary closing conditions.Promptfoo’s team, including co-founder and chief executive Ian Webster and co-founder Michael D’Angelo, will join OpenAI as part of the acquisition.Security tools for enterprise AIOpenAI said Promptfoo’s technology will help enterprises identify vulnerabilities in AI systems earlier in the development process and manage risks once AI agents are deployed in real-world workflows.The company added that the startup’s technology will be integrated directly into the Frontier platform for building and operating AI coworkers.As businesses begin deploying AI agents across operational systems and data environments, the company said evaluation, security and compliance are becoming critical requirements. Organisations need systematic methods to test agent behaviour, detect risks before deployment and maintain records to support governance and accountability.Red-teaming and testing capabilitiesPromptfoo develops open-source tools that allow companies to test AI systems and simulate attacks to uncover weaknesses, a practice known as red-teaming. Its tools are used to evaluate prompt behaviour and compare the performance of large language models such as ChatGPT, Anthropic’s Claude and Google’s Gemini.The San Francisco-based startup has about 11 employees and counts roughly a quarter of Fortune 500 companies among its customers. Promptfoo has raised $22.68 million in total funding and secured a $18.4 million Series A round in July last year led by Insight Partners, with participation from Andreessen Horowitz.Security features within FrontierFollowing the integration, Frontier will include automated security testing and red-teaming capabilities to help companies detect risks such as prompt injections, jailbreaks, data leaks, misuse of tools and policy violations by AI agents.The platform will also introduce monitoring and reporting features designed to help enterprises track testing, document changes over time and meet governance, risk and compliance requirements.“Promptfoo brings deep engineering expertise in evaluating, securing, and testing AI systems at enterprise scale,” said Srinivas Narayanan, chief technology officer of B2B Applications at OpenAI. He added that the company is “excited to bring these capabilities directly into Frontier.”Growing competition in AI securityThe deal comes as technology companies race to build more advanced AI agents capable of handling complex tasks independently while ensuring safeguards against misuse.Cybersecurity startups are increasingly developing AI-driven tools to defend against hackers even as malicious actors adopt similar technologies to probe vulnerabilities.Last week, OpenAI introduced an AI agent designed to help security teams identify and patch vulnerabilities in large databases, similar to tools offered by rival Anthropic.Promptfoo’s open-source project will continue to be developed even as its technology becomes part of Frontier.Webster said joining OpenAI would help accelerate efforts to secure AI systems as agents become more integrated with enterprise data and infrastructure.“As AI agents become more connected to real data and systems, securing and validating them is more challenging and important than ever,” Webster said. “Joining OpenAI lets us accelerate this work, bringing stronger security, safety, and governance capabilities to the teams building real-world AI systems.”First Published on March 10, 2026, 14:57:07 IST
