Zenity, a platform specializing in security and governance for AI agents, has announced a significant expansion of its AI security platform. The update introduces an intelligence layer for correlating AI-driven security incidents, extends coverage to agentic browsers across enterprises, and debuts a new open-source tool developed by Zenity Labs to evaluate emerging large language model (LLM) manipulation techniques.
According to Ben Kliger, co-founder and CEO of Zenity, the new release provides security teams with enhanced visibility into the intent behind AI agent actions. As organizations increasingly adopt AI agents, AI assistants, and agentic browsers, security teams have faced challenges in understanding how security incidents unfold across various identities, workflows, and environments. Traditional alerting often provides isolated signals without a coherent narrative.
Zenity’s latest advancements aim to offer a unified approach for detecting, analyzing, and governing AI behaviors in real-world enterprise settings. Kliger stated that their new Correlation Agent does not merely detect signals but interprets them, connecting data points and insights throughout the agent lifecycle into a single, coherent story. He added that this transforms scattered signals into high-confidence security narratives, aiming to eliminate guesswork, accelerate investigations, and provide clarity for operating AI safely at scale.
Key among the new capabilities is Zenity’s “Issues” feature, which correlates posture findings, runtime anomalies, identity relationships, and graph-based insights into high-confidence security incidents. This system unifies signals into narratives that explain the incident, its cause, and its impact, providing immediate visibility for security teams to begin investigations without manual reconstruction of events. The Correlation Agent is designed to capture intent, interpret behavior, surface manipulation attempts, and explain agent actions, reducing guesswork and investigation time.
Zenity is also expanding its AI security coverage to agentic browsers, initially focusing on ChatGPT Atlas, Perplexity Comet, and Dia. These tools represent a new source of “shadow AI,” as they autonomously read content across authenticated sessions and perform actions on behalf of the user. This creates a high-risk surface where a single malicious instruction could lead to data loss or credential misuse, with security teams often lacking visibility into the distinction between human and agent activity. Through Zenity’s device agent, organizations can discover these agentic browsers, monitor autonomous activity, apply data loss prevention, and detect intention-driven anomalies in real time, ensuring consistent policy application across all copilots and enterprise AI agents.
Furthermore, Zenity Labs is releasing an open-source tool named Safe Harbor, based on ongoing research into data structure injection and structured self-modeling attacks against LLMs. Safe Harbor allows an AI agent to call a dedicated safe action when it identifies something harmful, enabling it to pivot away from unsafe workflows rather than completing a malicious instruction. This tool is intended to help developers reduce the risk of runtime exploitations during the build process. Zenity aims to strengthen its position as a security platform built for how AI behaves across agents, browsers, workflows, and autonomous decision chains, helping organizations reduce noise, improve investigative depth, and gain visibility into the expanding ecosystem of agentic and AI-driven applications.
Zenity is a security and governance platform specifically designed for AI agents across SaaS, homegrown cloud platforms, and end-user devices. The company supports Fortune 500 enterprises in adopting AI securely by providing defense-in-depth with full-lifecycle coverage, including agent discovery, posture management, real-time detection, inline prevention, and response. Its agent-centric approach focuses on agent behavior, access, and tool invocation to eliminate blind spots and enforce consistent policies, enabling organizations to innovate with AI while maintaining security.
