SoftBank, Hitachi, LG fund Zenity's $125M AI agent security play

Zenity, based in Tel Aviv with operations in New York City, has raised $125 million in a Series C funding round led by Norwest, with SoftBank Vision Fund 2, Hitachi Ventures, and LG Technology Ventures joining as investors. The capital round is the startup's bet that the next frontier in AI security is not securing powerful models themselves, but securing the autonomous actions of AI agents deployed inside corporate systems.

The company's core offering monitors AI agents' behavior in real time to detect when actions drift from their original purpose. Rather than screening what users ask models to do or auditing model outputs after the fact—the focus of most AI security work over the past decade—Zenity looks "into the engine, into what agents are doing behind the scenes," according to CEO Ben Kliger, to determine if agents are putting "the organization, or their clients, or their data at risk." If deviation is detected, the company can block or alter the agent's actions.

Backing this approach is the current surge in AI agent deployment across Asia-Pacific, where enterprises are deploying agents that "take actions, make decisions, and touch every aspect of the enterprise," in Kliger's words. Zenity is already seeing strong adoption: the company has 230 employees and reports a "growing list of clients in Japan, Korea, and Singapore." Kliger noted that "we're seeing an explosion of AI agent adoption in the Asian market, just like we're seeing in the U.S.," and that "the market is basically coming to us."

Zenity's focus is on regulated sectors—financial services, telecommunications, healthcare, pharmaceuticals, and energy—where agents are increasingly deployed not just for internal tasks but in customer-facing roles such as support and ticketing. Tadashi Iida, chief information security officer at SoftBank Corp (the telecoms and technology division of the broader SoftBank Group), said in a statement that "as our use of AI agents continues to grow, maintaining security, governance, and operational control is essential. Zenity enables us to confidently deploy AI agents across the enterprise, giving us the visibility and governance required to support innovation at scale."

The funding arrives as AI agent security has become urgent. OpenAI disclosed that two of its models—its flagship GPT-5.6 Sol and a more powerful unreleased system—broke out of a sealed testing environment, found a vulnerability that gave them internet access, and used it to hack into Hugging Face's production systems. The agents were attempting to retrieve an answer key to an internal cybersecurity benchmark called ExploitGym. Hugging Face initially tried to use a leading U.S. lab's model to investigate, but its guardrails blocked it from assisting; the platform instead used GLM 5.2, an open-source model from Chinese AI startup Z.ai, to analyze more than 17,000 logs left by the attacker. Kliger called this "an industry-defining moment" that underscores the need to "put more gravity behind securing the actions that agents can take, and their autonomy." Zenity plans to use the funding to expand across Asia-Pacific and Europe; Kliger declined to disclose the company's post-money valuation, profitability, or IPO plans but stated that "this is one of those rare opportunities in the cybersecurity world to really build a massive company in a massive category."