Israeli startup Zenity bags $125M in funding to build the security layer for AI agents
SiliconANGLE Mike Wheatley ● Covered by 3 sources
Israeli startup Zenity just raised $125M to police AI agents running loose inside big companies. As firms let AI agents actually do work, someone has to stop them going rogue — that's now a $185M business.
Zenity just closed a $125 million Series C, led by Norwest with SoftBank Vision Fund 2, Qumra Capital, Intel Capital and a handful of others piling in. That brings the Tel Aviv-based startup's total raised to $185 million, and it's a bet on a problem most companies haven't fully grasped yet: AI agents that can actually do things — book orders, move data, call third-party tools — are a completely different security headache than chatbots that just answer questions.
Most of the AI security industry to date has obsessed over the model itself or the prompt window, trying to stop someone from tricking a chatbot into saying something embarrassing. Zenity's bet is that this misses the real risk. Once an agent has permission to touch enterprise systems and execute business processes on a human's behalf, the attack surface explodes into something identity and data security tools were never built to handle. CEO Ben Kliger put it bluntly to Calcalist: every organization is now either buying off-the-shelf agents or building its own, and securing what those agents actually do has become non-negotiable.
The platform works by mapping where agents live inside a company's systems, figuring out what each one is supposed to be doing, and then watching in real time to catch it doing something else. Rather than scanning prompts after the fact, it tries to judge intent before an action fires, giving security teams the option to approve, tweak or kill an agent's move on the spot — including shutting one down instantly if it looks hijacked or misused. It already plugs into Gemini, ChatGPT Enterprise, Microsoft Copilot, Claude, Cursor and custom in-house agents, which is a fairly wide net for a three-year-old company.
What's notable is that one of its investors, SoftBank, is also a customer — CISO Tadashi Iida said the visibility and control Zenity provides is what lets SoftBank deploy agents at scale without losing its nerve. And Zenity's credibility isn't purely commercial: its research arm, Zenity Labs, has publicly exposed serious flaws in Microsoft Copilot Studio and Perplexity's Comet browser, findings that feed straight back into the product. Analyst Holger Mueller framed the raise as a signal that the broader agentic AI market is maturing past the demo phase, with security spending following close behind. The new money will go toward scaling the platform, growing Zenity Labs, and pushing into Europe, Asia-Pacific and the Middle East.
My take
Every AI hype cycle eventually produces a security company that gets rich cleaning up the mess the hype cycle created, and Zenity is timing that perfectly. The real story here isn't the $125 million, it's that enterprises rushed autonomous agents into production before anyone worked out how to stop them from doing something stupid with real permissions — which is exactly the kind of sequencing that tends to end in a breach headline within a year or two.
Read more about this at: SiliconANGLE
