What Happened
Pluto Security, an Israeli-American startup founded by CEO Shahar Bahat, has raised $42 million to help enterprises secure the chaotic adoption of AI tools across their workforces. The funding comprises an $8.5 million seed round led by TLV Partners (with participation from Armis founders) and a $34 million Series A led by Jump Capital, joined by Skinos (Shlomo Kramer's venture firm), Ansa Capital, and angel investors.
The company launched its product in April 2026 and has already reached $2 million in annual recurring revenue with more than 30 enterprise customers, including Oscar Health, Workato, Sixt, and AlphaSense. Pluto employs 40 people across the United States and Israel.
The Key Details
The problem is speed and scope. According to Bahat, a newly released AI tool can appear inside a company's environment within two hours. The risk profile is surprising: roughly 60% of the threats Pluto detects across its customer base come from "vibe-coded" applications—tools built by non-developers using natural language prompts rather than traditional code. Finance departments are the largest source of these risks, followed by marketing and legal teams. Software developers rank only fourth.
A concrete example: Pluto discovered a finance employee at a Fortune 500 company had built an application to manage customer information and churn analysis, but left it exposed to the public internet.
Pluto's approach differs from traditional security tools. Rather than simply monitoring user behavior, the platform tracks "agent behavior" in real time—monitoring AI tools, skills, extensions, Model Context Protocol (MCP) servers, and the applications and agents employees create. The system connects to endpoint security and device management platforms, deploys its own sensors, and collects prompts, commands, running processes, permissions, and historical activity. Lightweight models running on endpoints assess what an agent is attempting to do.
Real-time intervention is a key differentiator. If an AI agent tries to escalate permissions beyond what the user intended, Pluto can block the action or inject instructions into the session prompting the agent to request confirmation. "We can inject context back to the session, and this way I can change how the agent is working," Bahat explained.
Community-led growth. In April, Pluto launched Plutonium, a free platform that lets security professionals scan public AI skills, connectors, and add-ons for risks, alongside an MCP scanner and a catalog of verified components. Bahat says thousands of security practitioners use Plutonium, and the community now drives approximately 70% of the company's inbound sales pipeline.
Every proof-of-concept engagement has uncovered active threats. Bahat told TechCrunch: "There was not one POC in which we entered and didn't find malicious stuff... some of them had active live attacks while we did the POC."
Why It Matters
Enterprise AI adoption has outpaced security frameworks. Organizations that approved ChatGPT, Copilot, or custom GPTs for productivity gains now face a shadow AI landscape where sensitive data flows through unvetted tools, extensions, and agent workflows—often built by employees with no security training. Traditional data loss prevention (DLP) and endpoint detection tools were not designed to interpret the intent of autonomous AI agents or the chain of permissions they request.
Pluto's early traction suggests the market recognizes this gap. The $2 million ARR achieved in roughly six months, combined with a 70% inbound pipeline, indicates strong pull from security teams who have already encountered the problem.
The "vibe coding" phenomenon—non-technical staff building functional applications through prompts—democratizes software creation but bypasses every traditional control: code review, security testing, deployment pipelines, and access management. Finance teams handling customer data, marketing teams connecting CRM systems, and legal teams processing contracts are now de facto application developers without the safeguards.
What It Means for Readers
For security leaders: The category of "AI security posture management" is emerging rapidly. Pluto's agent-behavior monitoring, real-time intervention, and community intelligence (via Plutonium) represent one approach. Expect more vendors to address MCP server risks, skill/extension vetting, and agent permission governance.
For IT and procurement teams: Any AI tool adoption policy must now account for extensions, connectors, and user-built agents—not just the primary LLM interface. Vendor risk assessments should include the ecosystem of add-ons each platform enables.
For non-technical employees: The convenience of building tools with prompts carries hidden responsibility. An application created in minutes can expose customer data, violate compliance requirements, or become an entry point for attackers if left unsecured.
For African enterprises: As AI adoption accelerates across Nigerian and African businesses—particularly in fintech, banking, and professional services—the same dynamics apply. Finance teams using AI for reporting, marketing teams automating campaigns, and operations teams building internal tools will create similar shadow AI risk. Security budgets and expertise remain constrained in many markets, making automated, agent-aware monitoring especially valuable.
What Happens Next
Pluto plans to invest the new capital in product development, expanding its security research team, and growing its U.S. presence. The company will likely deepen its Plutonium community platform as a lead-generation engine and threat intelligence source. Competitors in the AI security space—including startups addressing prompt injection, data exfiltration, and agent governance—will continue raising large rounds, signaling sustained investor conviction that this is a distinct market category.
VANTIX Takeaway
The enterprise AI security problem is no longer theoretical. Within months of widespread AI tool deployment, organizations face a reality where non-technical employees are shipping live applications that touch sensitive data, often exposed to the internet, with no security review. Pluto's early success demonstrates that security teams are willing to pay for visibility and control over agent behavior—not just user behavior. For any organization deploying AI at scale, the question is not whether shadow AI exists, but whether you can see what your agents are doing right now.
