Onyx Security, the AI control company, announced a $113 million Series B funding round led by Bessemer Venture Partners, with participation from Cyberstarts, TCV, Conviction, FirstMark, Vintage, QuantumLight and G Squared. This brings the company’s total funding to $153M since its founding exactly two years ago.
Since coming out of stealth four months ago, Onyx has grown revenue fourfold, and in June, Anthropic announced its integration with Onyx to secure enterprise AI adoption. This comes on the heels of dozens of other AI providers being integrated with the Onyx platform.
In the past few months Onyx has seen the advent of a new type of models, such as Mythos and Fable, possessing such advanced capabilities they were initially deemed too dangerous for public release. With the advancement in model intelligence, enterprise AI adoption is entering a new phase of sophisticated AI agents that are incredibly capable, yet hard to control. Employees are giving these agents access to critical systems and tasking them to automate increasingly complex jobs. This presents a fundamental new risk: agents are empowered but not accountable for their actions.
Onyx’s secure AI control plane enables enterprises to discover and control how advanced AI is used in their environment, paving the way for further agent adoption. Using its proprietary models, Onyx is able to monitor each step of agent reasoning and to correct unintended or malicious agent actions. This gives security, governance, and infrastructure teams the assurance they need to support AI adoption across the company.
Companies have begun to recognize the urgency of the AI control challenge. “In 2025, less than one percent of actions were made by agents,” said Maxim Bar Kogan, CEO and Co-founder of Onyx. “These new autonomous actors will soon be taking the overwhelming majority of actions, which begs the question – how do we guarantee these actions are legitimate?”
Onyx is trusted by dozens of leading enterprises, including Fortune 500 companies across banking, technology, insurance, energy, and other critical industries, to provide control over third-party agents as well as home-grown agents built by internal engineering teams.
“Onyx gives us the visibility and control we need to enable safe use of coding agents across multiple service providers,” explained Uros Solar, Head of Security Engineering at Revolut. “The platform’s unique approach allows us to evaluate and stop threats at runtime.”
“AI agents are becoming an important part of how modern companies operate, but enterprises need trust, control, and accountability before they can adopt them at scale,” said Amit Karp, Partner at Bessemer Venture Partners. “Onyx is building the control layer that makes this possible. Maxim and Gil have identified one of the most urgent challenges in enterprise AI, and we believe that Onyx is positioned to define this category.”
“As AI agents become embedded in critical business workflows, enterprises need a way to ensure they operate safely, predictably, and within policy,” said Hila Zigman, General Partner at Cyberstarts. “Onyx is building the control layer that makes enterprise AI adoption possible at scale. We believe this will become one of the defining security categories of the coming decade.”
The new capital will be used for training the next generation of Onyx’s proprietary models and for scaling the enterprise go-to-market in the United States and beyond.
About Onyx Security
Onyx Security is the secure AI control plane for the enterprise. Its platform gives security, governance, and infrastructure teams unified control over AI agents, ensuring they operate securely, stay compliant, and deliver measurable business value. Powered by proprietary models and agentic technology, Onyx is trusted by leading Fortune 500 companies to bridge the gap between AI ambition and enterprise security. To learn more, visit onyx.security.
View source version on businesswire.com: https://www.businesswire.com/news/home/20260729713522/en/
Media gallery
