Israeli startup Zenity bags $125M in funding to build the security layer for AI agents
SiliconANGLE Mike Wheatley ● Covered by 3 sources
Zenity, an Israeli startup, just raised $125M to build security tools for AI agents. As companies let AI agents run real business tasks, someone has to watch them so they don't go rogue.
Based on reporting by SiliconANGLE, Mike Wheatley — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Enterprises are done experimenting with AI agents and are now putting them to work on real tasks, and that shift has created a security gap that nobody quite had a name for until Zenity started filling it. The Israeli startup just closed a $125 million Series C, led by Norwest with SoftBank Vision Fund 2, Qumra Capital, Hitachi Ventures, LG Technology Ventures, Vertex Ventures, DTCP and Intel Capital all chipping in. That brings Zenity's total raised to $185 million, a war chest the company plans to spend on scaling its platform and expanding its research arm.
Most AI security tools so far have focused on the model itself, or the prompt layer, trying to stop bad actors from tricking a chatbot into saying something it shouldn't. Zenity is chasing a different problem. Autonomous agents don't just talk, they act. They log into enterprise systems, call third-party tools, make decisions, and execute business processes on behalf of humans. That's a sprawling new attack surface that the identity and application security tools built for human users were never designed to cover.
Zenity's approach is to map out where agents live inside a company's computing environment, figure out what each one is actually supposed to do, and then watch it in real time. Rather than inspecting prompts after the fact or reacting once something's gone wrong, the platform tries to read the intent behind an action before it happens, letting security teams approve, tweak or block it on the spot. If an agent starts behaving outside its lane, Zenity can shut it down immediately. The system works across major agent environments including Gemini, ChatGPT Enterprise, Microsoft Copilot, Claude and Cursor, as well as agents companies build in-house.
CEO Ben Kliger told Calcalist that agent adoption is accelerating fast, with organizations either buying off-the-shelf agents or building their own, and that securing what those agents do is now core to making AI adoption work at all. Constellation Research analyst Holger Mueller sees the funding itself as a signal — money flowing into security around AI, rather than just into AI models, suggests the sector is maturing past the hype phase. He also flagged the obvious open question: whether Zenity can actually deliver on protecting against agents going rogue, something he expects to become clear within a few quarters.
What gives Zenity extra credibility here is Zenity Labs, its research arm, which has already surfaced notable vulnerabilities in systems like Microsoft's Copilot Studio and Perplexity's Comet browser. Kliger says that research loops directly back into the product. Tellingly, SoftBank is both an investor in this round and a customer, with its CISO Tdashi Iida saying the platform gives the company the governance and visibility it needs to deploy agents at scale. Zenity says it already counts customers across the Fortune 500 and Global 2000 in healthcare, finance, pharma, manufacturing and tech, and plans to push further into Europe, Asia-Pacific and the Middle East with the new funding.
My take — AI-written commentary, not fact-checked reporting
Every gold rush needs someone selling shovels, and AI agents running loose inside enterprise systems were always going to need a bouncer. What's notable is that the smart money is betting the real risk isn't the model hallucinating — it's the agent actually doing something, unsupervised, with real system access. Investors chasing the flashy model layer for the last few years might be kicking themselves for missing this angle; the boring plumbing of governance is where the money's heading now.
Read more about this at: SiliconANGLE