‹ Back

RunSybil Raises $40M for AI Cybersecurity Push

 

RunSybil, an AI cybersecurity startup founded by OpenAI’s first security hire, has secured $40 million in venture funding led by Khosla Ventures.

 

 

RunSybil uses AI agents to autonomously test company software for security vulnerabilities. The funding round also included investors S32, the Anthology Fund from Anthropic, Menlo Ventures, Conviction, Elad Gil, and angel investors including Nikesh Arora, Amit Agarwal, Jeff Dean, and other founders from OpenAI, Palo Alto Networks, Stripe, and Google.

 

The company did not disclose its valuation in the new funding round. Its AI agent, Sybil, conducts continuous autonomous penetration tests against live applications, identifying and documenting real security weaknesses without human intervention, differentiating it from tools that analyze source code prior to deployment.

 

RunSybil tests live systems the way hackers do, exploring systems, chaining vulnerabilities, and probing authentication boundaries to find paths to sensitive data.

 

Automating ‘ethical hacking’

 

Companies traditionally rely on penetration tests, bug bounty programs, and internal red teams to simulate cyberattacks. RunSybil’s AI system automates much of this process, continuously probing applications for vulnerabilities as new code is deployed.

 

The company argues automation is necessary as AI reshapes business operations across procurement, legal, finance, engineering, and operations. Security testing remains scheduled separately, which can be challenging for regulated industries like finance, insurance, and healthcare.

 

RunSybil was co-founded in 2023 by Ari Herbert-Voss, OpenAI’s first security hire, and Vlad Ionescu, former leader of offensive security red teams at Meta. The founders combine expertise in building advanced AI systems and ethical hacking.

 

Herbert-Voss emphasized that the system transforms how and when customers detect and fix security issues, embedding capabilities into software development rather than treating security as a one-time project.

 

‘On the edge’ of the AI security frontier

 

Vinod Khosla, who invested early in OpenAI, told Fortune that RunSybil operates on the technological frontier of AI security. He noted that few competitors exist in this segment, though security incumbents like Palo Alto Networks may enter over time.

 

Khosla stressed concern about AI cyber capabilities being misused by adversaries and highlighted that RunSybil’s founders are building tools that will be essential as software complexity and AI-driven development increase.

 

Herbert-Voss’ background blends hacking and artificial intelligence (AI). Raised in Utah, he participated in online hacker communities before pursuing a Ph.D. at Harvard in machine learning. He left Harvard after recognizing AI’s potential to rapidly scale capabilities and joined OpenAI to perform security research.

 

Evolving cyber capabilities with LLMs

 

After observing the potential of GPT-2, Herbert-Voss recognized how AI could impact cyber operations. He shared demonstrations with OpenAI CEO Sam Altman and policy head Jack Clark, who expressed concern about misuse and recruited him for security research.

 

By 2022, Herbert-Voss saw how offensive cyber capabilities could evolve with accessible language models, motivating him to start RunSybil as a research project.

 

RunSybil works with startups including Cursor, Turbopuffer, Notion, Baseten, and Thinking Machines Lab, as well as undisclosed major financial institutions and Fortune 500 companies. Customers report discovering critical vulnerabilities that traditional methods missed.