RunSybil Raises $40 Million to Automate Cybersecurity with AI
AI-powered cybersecurity startup RunSybil has secured $40 million in a funding round led by Khosla Ventures, with participation from S32, Anthropic’s Anthology Fund, Menlo Ventures, Conviction, and Elad Gil. Angel investors include prominent figures from OpenAI, Palo Alto Networks, Stripe, and Google.
Automated Penetration Testing
Founded in 2023 by Ari Herbert-Voss, OpenAI’s first security researcher, and Vlad Ionescu, former Head of Offensive Security Red Teaming at Meta, RunSybil utilizes an AI agent named Sybil to conduct continuous, autonomous penetration tests on live applications. This approach differs from traditional security tools like Claude Code Security, which analyzes source code for known vulnerabilities before deployment. Instead, Sybil actively probes running software, mimicking the tactics of a hacker to identify and exploit security weaknesses.
The Rise of AI-Driven Security
Traditional cybersecurity strategies often rely on manual penetration tests, bug bounty programs, and internal “red teams.” RunSybil aims to automate these processes, continuously testing applications as new code is released. This automation is increasingly crucial as companies integrate AI across all operations – including procurement, legal, finance, engineering, and operations – and as AI agents become more prevalent.
Addressing a Critical Gap in Cybersecurity
The company’s technology addresses a growing need for continuous security assessment, particularly in highly regulated industries like finance, insurance, and healthcare, where strict compliance requirements are paramount. RunSybil’s founders believe their platform provides the necessary documentation and capabilities to meet these rigorous standards.
A Frontier Investment for Khosla Ventures
Vinod Khosla of Khosla Ventures, an early investor in OpenAI, sees RunSybil as a leader in a nascent field. He emphasized the lack of competition in automating offensive security and highlighted the unique expertise of Herbert-Voss, stating he possesses knowledge “nobody’s really knowledgeable about except individuals like [Herbert-Voss].” Khosla similarly expressed concern about the potential for AI’s cyber capabilities to be exploited by adversaries.
From Hacker Roots to AI Security
Ari Herbert-Voss’s journey began with an interest in the online hacker community. After pursuing a Ph.D. At Harvard University in machine learning, he joined OpenAI in 2019, becoming their first security researcher. He left OpenAI in 2022, recognizing the rapidly evolving offensive cyber capabilities enabled by large language models (LLMs) and the need to proactively address potential threats.
Current Clients and Future Growth
RunSybil currently serves a diverse client base, including startups like Cursor, Turbopuffer, Notion, Baseten, and Thinking Machines Lab, as well as major financial institutions and Fortune 500 companies. Customers have reportedly discovered critical vulnerabilities that went undetected by traditional security methods.
Source: Fortune