Summary
RunSybil, a new cybersecurity company, has raised $40 million in a recent funding round. The startup uses artificial intelligence to act like a hacker and find weak spots in a company's software. This process, known as "ethical hacking," helps businesses fix security problems before real criminals can find them. The company was started by experts who previously worked at OpenAI and Meta.
Main Impact
The rise of AI is changing how companies build software, but it is also changing how hackers attack. RunSybil is trying to stay ahead of these threats by automating the way companies test their defenses. Instead of hiring human experts to check for holes once or twice a year, RunSybil’s AI agent works all the time. This shift allows businesses to find and fix bugs as soon as they update their software, making it much harder for cybercriminals to succeed.
Key Details
What Happened
RunSybil created an AI agent called Sybil. This tool does not just look at the written code of an application. Instead, it interacts with software that is already running. It tries to break into systems, bypass login screens, and reach private data just like a real-world attacker would. By doing this, it provides a realistic view of how safe a company actually is. This is different from many other security tools that only look for known errors in the code before the software is even turned on.
Important Numbers and Facts
The $40 million funding round was led by Khosla Ventures. Other big names joined in, including the Anthology Fund from Anthropic and Menlo Ventures. Several leaders from major tech companies like Google, Stripe, and Palo Alto Networks also invested their own money. RunSybil was started in 2023 by Ari Herbert-Voss and Vlad Ionescu. Herbert-Voss was the very first person hired to handle security research at OpenAI, while Ionescu led teams at Meta that specialized in attacking their own systems to find flaws.
Background and Context
For a long time, companies have used "penetration tests" to stay safe. This usually involves hiring a team of experts to try and break into their network. While this works, it is often slow and expensive. Because software is now updated almost every day, a security test done in January might be out of date by February. RunSybil aims to solve this by making the testing process continuous.
The founders bring a unique set of skills to this problem. Ari Herbert-Voss grew up interested in the hacker community before studying machine learning at Harvard. He joined OpenAI early on because he realized that large AI models would change the world of cybercrime. He saw that AI could make it much easier for bad actors to launch attacks, so he decided to build a tool that uses that same power for defense.
Public or Industry Reaction
Investors are very excited about this new approach. Vinod Khosla, a well-known investor who was an early supporter of OpenAI, believes RunSybil is working on the most advanced edge of technology. He noted that there are very few people in the world who understand both high-level AI and complex hacking techniques. Khosla also mentioned that he is worried about AI hacking tools being used by foreign governments, which makes domestic defense tools even more important.
Several fast-growing startups, such as Notion and Cursor, are already using RunSybil. The company also says it works with large banks and Fortune 500 firms, though it has not shared all of their names yet. These early customers report that the AI has found serious security issues that traditional tools missed.
What This Means Going Forward
As more parts of a business—like legal, finance, and engineering—start using AI, the risk of a cyberattack grows. Security can no longer be a separate task that happens once in a while. It must be built into the way software is made. RunSybil plans to use its new funding to grow its team and improve its AI agent. The goal is to make "autonomous hacking" a standard part of how every major company protects its data. In the future, we may see more competition in this space as older security companies try to build their own AI agents.
Final Take
The battle between hackers and security teams is moving faster than ever because of artificial intelligence. By using AI to think like an attacker, RunSybil gives companies a way to find their own weaknesses before someone else does. This funding shows that the industry believes automated, "offensive" security is the best way to keep data safe in the modern age.
Frequently Asked Questions
What is an AI penetration test?
It is a security check where an AI program acts like a hacker to find weaknesses in a company's live software. It tries to break in to show the company where they need to improve their defenses.
How is RunSybil different from other security tools?
Most tools just scan the written code for mistakes. RunSybil actually attacks the software while it is running, which helps find more complex problems that code scanners might miss.
Who started RunSybil?
The company was started by Ari Herbert-Voss, who was the first security researcher at OpenAI, and Vlad Ionescu, who led security attack teams at Meta.