Autonomous AI agents are now doing real work inside the enterprise. They write and ship code, summarize board materials, query production databases, and call external tools on a user’s behalf. Each of those actions moves data, and most security teams cannot see a single one.
That blind spot is why Google Cloud convened the Security Innovations Forum 2026 today in Sunnyvale, CA. Cyberhaven is proud to join as a partner at this invite-only event, bringing together 200 chief information security officers (CISOs) and security executives. Alongside more than a dozen independent software vendors, our team of experts will demo live at our partner kiosk.
Enterprises are building with Google’s AI in two places at once: developers working in Google Antigravity and Gemini command line interface (CLI), and business users running agents in Google Cloud Gemini Enterprise and across Google Workspace. The Security Innovations Forum will showcase production-ready AI agents executing high-stakes tasks in real time. Google Cloud will show how it shifts the enterprise from a reactive posture to continuous, autonomous defense with Google AI Threat Defense, alongside partner announcements and demos. With Cyberhaven, we’ll show you how the world’s most innovative companies are securely embracing the new world of work.
Meet Linea AI, Cyberhaven’s Data Security Agent for the Agentic Enterprise
Cyberhaven is a leader in data security for the agentic enterprise. Many of the world’s most innovative companies – frontier AI labs, robotics manufacturers, pioneers in space and defense, the top quant hedge funds, breakthrough biotech firms, major banks, oil & gas companies, and law firms – trust Cyberhaven to protect their most sensitive data across human and agentic workflows. That’s because Cyberhaven traces the full lifecycle of their data, adapting protection to changing context.
We’re excited to join Google Cloud’s Security Innovations Forum today because Cyberhaven was an early innovator in agentic AI, including using production-grade AI agents to secure data. In 2023, we introduced some of the first AI agents to autonomously detect and investigate data security incidents. Using our Large Lineage Model, a proprietary AI model based on data lineage to help agents secure workflows at scale, Cyberhaven’s AI agents help data security teams work more efficiently and effectively.
Recently, we expanded our agentic capabilities in Cyberhaven Flow, our all-new AI-native data security solution built around agentic AI: a team of agents and subagents called Linea AI. Flow adds configuration agents, classification agents, and more to our detection and analysis agents, putting agentic AI at the heart of data security. Linea agents are built on the Gemini Agent platform
To give you some simple examples, here’s how Cyberhaven’s Linea AI helps security teams automate work and improve quality on tasks they do by hand today:
- Classifies sensitive data across endpoints, browsers, SaaS applications, and AI agents.
- Turns plain-language intent into datasets and policies.
- Reviews every event, not only policy violations, so it catches risk no rule anticipated.
- Investigates the incidents that matter, gathering lineage evidence and recommending the response.
Linea AI runs on Google’s Gemini models and on Cyberhaven's Large Lineage Model, and it extends existing data policies to agents built on Google Antigravity and Gemini Enterprise.
How Cyberhaven’s Linea AI drives better business outcomes
As the use of AI agents increases, this creates a major scaling challenge for security teams. Security programs that wait for people to classify, tune, triage, and investigate cannot keep pace with agents moving data at machine speed. Cyberhaven’s Linea AI solves that, meeting the agentic challenge with agentic solutions. Here is what that looks like in three scenarios.
Securing Developers on Antigravity
A developer asks Antigravity CLI to refactor a service. The agent reads a configuration file holding production credentials and a customer data export, creates a repository under the developer's personal GitHub account, and pushes the code. Cyberhaven reconstructs the session and classifies what moved through it. The customer's existing policy on source code and credentials fires with the full data-flow timeline attached, and the transcript shows whether the developer asked for it or a poisoned README did.
Securing Business Users on Gemini Enterprise and Google Workspace
A finance executive uses ‘Ask Gemini’ in Google Slides to summarize board-level results, then copies the summary into personal cloud storage. That transformation defeats pattern-matching data loss prevention (DLP). Cyberhaven captures the Gemini interaction through the browser, classifies the content as confidential financial data, and decorates the summary with the lineage of the board deck behind it. The existing financial data policy fires, and Cyberhaven distinguishes the corporate Google Workspace account from a personal Gemini account.
Shadow MCP and Supply-Chain Exposure
A developer's Gemini CLI loads a third-party Model Context Protocol (MCP) server that, like the postmark-mcp incident, quietly forwards content to an external endpoint. Cyberhaven's inventory shows the server in use, on which endpoints, by whom, and how often, so when a disclosure breaks, security answers "are we affected?" in under a minute. Data routed to the unsanctioned server stops even when a poisoned tool description, not the user, initiated it.
Learn more about how Cyberhaven is securing data for the Agentic Enterprise
If you’re attending the Security Innovations Forum today in Sunnyvale, stop by the Cyberhaven kiosk to see a live demo of Cyberhaven Flow and Linea AI agents on Antigravity and Gemini Enterprise.
Can’t make it to Sunnyvale? Schedule time with us to learn more.
About Cyberhaven
Cyberhaven is a leader in Data Security for the Agentic Enterprise, trusted by the world's most innovative companies to protect their most sensitive data. Cyberhaven traces the full lifecycle of your data, adapting protection as context changes. Cyberhaven protects workflows, not just data, by connecting lineage, identity, and behavior to understand how data is created, copied, fragmented, and shared across workflows. Operating in real time where humans and agents act on data, Cyberhaven helps stop risk before it becomes loss. AI changed work. Cyberhaven protects it.
Learn more at cyberhaven.com
.avif)
.avif)
