RunSybil, a cybersecurity startup that uses AI agents to carry out penetration tests, raised $40m in a Series A round announced on 18 March 2026. Khosla Ventures led the round, according to Fortune and SiliconANGLE. Valuation was not disclosed.
Other investors were S32, the Anthology Fund (a fund backed by Anthropic), Menlo Ventures, Conviction and Elad Gil. Fortune also named angel investors including Nikesh Arora, Amit Agarwal and Jeff Dean, along with founders and leaders from OpenAI, Palo Alto Networks, Stripe and Google.
“We invest in founders who tackle large, unsolved problems with technically ambitious solutions.”Vinod Khosla, Founder, Khosla Ventures
What Sybil does
RunSybil's product is an AI agent called Sybil that runs continuous, autonomous penetration tests on live applications. According to Fortune, it finds, exploits and documents real vulnerabilities without people in the loop. It is aimed at the work usually done through scheduled penetration tests, bug bounties and red-team exercises, and probes applications as new code ships.
Fortune contrasted this with tools that analyse source code before deployment. Sybil tests software that is already running, chains together vulnerabilities and probes authentication boundaries. SiliconANGLE reported that it needs no access to source code, and that the company says it cut false positives by 90% compared with standard security scanners, a figure the company itself supplied.
Chief executive Ari Herbert-Voss told SiliconANGLE: "We're the first to provide comprehensive black-box testing using AI to reason like a security researcher."
Customers and founders
Fortune listed Cursor, Turbopuffer, Notion, Baseten and Thinking Machines Lab as customers. It said RunSybil also works with several major financial institutions and Fortune 500 companies that it declined to name. The company says continuous testing matters most in regulated sectors such as finance, insurance and health care.
RunSybil was founded in 2023 by Herbert-Voss and Vlad Ionescu. Herbert-Voss joined OpenAI in 2019 as its first security research hire and left in 2022 to start the company, concerned that language models could quickly expand offensive cyber capabilities. Ionescu previously led offensive security red teams at Meta. Neither report gave a headcount or a headquarters city.
Why Khosla backed it
Vinod Khosla, founder of Khosla Ventures, said: "We invest in founders who tackle large, unsolved problems with technically ambitious solutions." Fortune quoted him calling the work of adding security and penetration testing to the AI world frontier, and saying few people are knowledgeable about it. He also said incumbents such as Palo Alto Networks might eventually enter the niche.
What the money is for
SiliconANGLE reported that the money will go to research and development, to expanding agentic security testing, to scaling go-to-market teams and to hiring more researchers. A third report, from Built In, said the company was hiring for engineering, research and customer-facing roles.
The company has not said how many customers it has or what it earns from them, and prior funding was not reported in the coverage reviewed.
Company website: runsybil.com