Pluto Security Raises $42M to Monitor Enterprise AI Risk

Pluto Security is closing a $42 million funding round as enterprises struggle to contain security risks created by employee-driven AI adoption. The startup, which emerged from stealth with this announcement, joins a growing number of companies built on the conviction that traditional security software cannot adequately address risks specific to AI agents.

The funding comprises an $8.5 million seed round led by TLV Partners, with participation from the founders of cybersecurity firm Armis, alongside a $34 million Series A led by Jump Capital. Additional backers include Skinos, the venture firm founded by Shlomo Kramer who co-founded Check Point, as well as Ansa Capital and other angel investors.

Pluto began selling its platform to enterprises in April, meaning the startup has been operating for less than seven months at the time of this announcement. Despite this short timeline, the company has already surpassed $2 million in annual recurring revenue and accumulated more than 30 enterprise customers. Those customers include Oscar Health, Workato, Sixt, and AlphaSense, spanning healthcare, automation, car rental, and financial intelligence sectors.

Shahar Bahat, CEO and co-founder, built the company around a specific observation about how AI tools spread through enterprise environments and create security problems that escape the view of traditional monitoring systems.

Where Enterprise AI Risks Actually Originate

The core problem Pluto addresses is that employees outside engineering departments are now building applications using AI tools, often without the security oversight that developer-created systems receive. According to Bahat, approximately 60 percent of the security risks Pluto detects across its customer base come from tools built by non-developers.

Finance departments are the largest source of such risks, Bahat said, followed by marketing and then legal teams. Developers, who typically work within more structured security frameworks and code review processes, rank fourth in terms of risk generation. The gap exists because non-technical departments encounter problems in their daily work—managing customer data, analyzing market trends, processing contracts—and turn to AI tools as quick solutions. Those solutions often bypass the security controls established for traditional enterprise software.

A Fortune 500 Example

Bahat cited a concrete example from her company’s work with a Fortune 500 corporation. A finance employee created an AI-powered application to manage customer information and detect customer churn. The application worked as intended but was left exposed to the internet without any access controls. Only when Pluto’s platform scanned the environment did the vulnerability become apparent.

This pattern repeats across Pluto’s customer base. In fact, Bahat stated that during every proof-of-concept evaluation the company has conducted, it found at least one instance of malicious activity or dangerous misconfigurations. In some cases, those evaluations uncovered active attacks already occurring within the customer’s environment while Pluto was conducting its assessment.

The Speed Problem

Enterprise environments now face another complication: AI applications can appear within a company’s network within just two hours of their public release. An employee discovers a new tool, tries it out, and integrates it into their workflow before traditional security reviews can even begin. This speed differential means that by the time a company’s security team learns about a tool, it may already be operating on sensitive systems and handling confidential data.

The Monitoring Approach: Agent Behavior, Not User Behavior

Read also: Meta Releases Muse Wonder, Its First Frontier AI Model · Former Plaud exec raises $26M for always-on AI listening assistant · Apple Hires Huxe Team, Licenses Podcast AI Without Full Acquisition · Mosqitter transforms mosquito traps into disease surveillance network

Pluto’s core insight is that AI creates a new category of security problem that traditional tools were never designed to address. Rather than viewing AI tool usage as a user behavior issue—how an employee acts—the platform treats it as an agent behavior issue—what an AI agent does when operating on behalf of that user.

Bahat described this shift in approach: “We’re moving from user behavior to agent behavior.”

The platform monitors multiple layers of the AI ecosystem within an organization. It tracks the AI tools employees use, the skills and extensions integrated into those tools, MCP servers that provide functionality, and the applications or agents that employees create with AI. Pluto collects data through connections to endpoint security and device management systems, where it can deploy sensors to gather information about prompts, running processes, executed commands, system permissions, and historical user activity.

Intervention During Operation

What distinguishes Pluto from other security approaches is its capacity to intervene while an agent is actively working. If an AI agent attempts to escalate its permissions beyond what the user intended, Pluto’s system can either halt the action entirely or inject new instructions into the session that prompt the agent to seek confirmation before proceeding.

“We can inject context back to the session, and this way I can change how the agent is working,” Bahat explained. This real-time intervention goes beyond detecting problems after they occur. Instead, Pluto prevents certain actions from completing by reshaping how the agent behaves mid-execution.

Lightweight Analysis on Endpoints

To accomplish this monitoring without sending sensitive data to external systems, Pluto deploys its own “lightweight” models directly on employee devices. These on-device models analyze what an agent intends to do and assess risk levels without requiring full prompts or sensitive information to leave the endpoint.

How Pluto Builds Its Customer Pipeline

The startup has built an unconventional go-to-market strategy that extends beyond direct enterprise sales. In April, concurrent with launching its commercial platform, Pluto released Plutonium—a free platform where security professionals can scan publicly available AI skills, connectors, and add-ons for vulnerabilities. The platform includes an MCP scanner and a catalog of components that have been verified against known risks.

Plutonium has attracted thousands of security professionals and has become embedded in how some practitioners assess AI tools before deploying them. Bahat said that 70 percent of Pluto’s current sales pipeline originates from inbound interest, though she did not break down the specific proportion that comes directly through Plutonium users. The free tool serves multiple purposes: it provides valuable security information to the broader community, generates awareness of Pluto’s brand and approach, and creates a natural funnel toward the company’s paid platform for enterprises that need continuous monitoring and enforcement.

Investing for Scale

Pluto plans to allocate its new capital toward three areas: developing its product further, expanding its security research efforts, and growing its presence in the United States market. The company currently operates with 40 employees across offices in the United States and Israel. The funding suggests that investors believe AI security will become a substantial market category, with Pluto well-positioned to capture a meaningful share based on its early traction and differentiated approach to agent monitoring and intervention.

Source: TechCrunch