AivexaNewsSearch
AI news for builders and product teamsChecked every hour

Pluto Security raises $42M as AI gives enterprise security teams a new headache

Collected Oct 13, 2026

Pluto Security came out of stealth with $42 million in funding, positioning itself among a new crop of security startups built to address the risks that arrive when employees across a company get access to AI tools. The company's chief executive and co-founder, Shahar Bahat, told TechCrunch that an AI tool can appear inside a corporate environment as little as two hours after it is released.

The funding breaks into two rounds. An $8.5 million seed round was led by TLV Partners, with the founders of the cybersecurity firm Armis also investing. The company then raised a $34 million Series A led by Jump Capital, with participation from Skinos, the venture firm of Check Point co-founder Shlomo Kramer, along with Ansa Capital and other angel investors.

Pluto's commercial traction is modest but real. Bahat said the company started selling in April and has crossed $2 million in annual recurring revenue, with more than 30 enterprise customers including Oscar Health, Workato, Sixt, and AlphaSense. It employs 40 people across the United States and Israel, and plans to put the new capital into product development, security research, and expanding its U.S. presence.

What Pluto sells is a way for companies to gain context, risk analysis, visibility, and control over how employees use AI tools. The product monitors the AI tools workers adopt, the skills, extensions, MCP servers, and other components those tools depend on, and the apps and agents being built internally. It identifies what it considers risky components, watches for dangerous agent activity, and enforces security policies without blocking employees from using or creating AI tools. For a developer, the interesting piece is that Pluto connects to endpoint security and device management systems, deploying sensors through those systems or directly onto devices, and gathers prompts, commands, running processes, permissions, and prior user activity. It runs its own lightweight models on endpoints to judge what an agent is trying to do.

That endpoint placement matters. It is what allows the system to intervene while an agent is mid-task. If an agent tries to take permissions beyond what the user intended, the system can stop the action or inject instructions into the session that push the agent to ask for confirmation. Bahat described the approach as injecting context back into the session to change how the agent works. The pitch behind it is that AI security requires watching what agents do in real time, not just tracking what employees build. Bahat framed the shift as moving from user behavior to agent behavior.

The risk profile Pluto describes is lopsided in a way that may surprise security teams. Bahat said roughly 60% of the risks the company detects across its customer base sit in tools vibe-coded by non-developers. Finance departments generate the most such risks, she said, followed by marketing and legal, with developers in fourth place. One example: a vibe-coded app built by a finance employee at a Fortune 500 company to manage customer information and churn, left exposed to the internet.

Bahat said risk varied with the mindset of the chief information security officer and how willing an organization was to adopt AI, but that the problems showed up regardless. She said every proof of concept Pluto entered turned up malicious material, whether extension packages or skills, and that some customers had active live attacks running while the POC was underway.

Go-to-market is unusual for an enterprise security vendor. In April, Pluto launched Plutonium, a free platform where anyone can check risks tied to publicly available AI skills, connectors, and add-ons, alongside an MCP scanner and a catalog of verified components. Bahat claims thousands of security professionals use Plutonium and that the community around it has become a source of feedback and prospective customers. She said 70% of the sales pipeline is now inbound, though she did not say how much of that arrives specifically through Plutonium.

Some context on the category: the enterprise AI security market has attracted a wave of young vendors betting that established security software, built around monitoring traditional applications and network traffic, is a square-peg-round-hole fit for AI tools, connectors, and usage. Pluto is making the same argument, with the added claim that agent sessions need to be steered rather than simply blocked. That stance raises questions a buyer would want answered: how much latency an on-endpoint model adds, what fraction of agent actions can be safely interrupted without breaking a workflow, and whether injected confirmation prompts scale across thousands of sessions. Bahat's numbers on detections come from her own POCs, which is typical for an early-stage vendor and worth treating as directional rather than benchmarked.

Why it matters: security teams are being asked to govern tools they never approved, built by employees who are not developers, and now agents that act on their own within those tools. For developers and product teams, the practical near-term effect is more instrumentation at the endpoint, more policy prompts interleaved into agent sessions, and a growing expectation that internal apps and agent permissions be inventoried and reviewable. The funding suggests investors expect that control layer to be a durable category rather than a temporary gap, and that inference is already being priced in.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

The problem with giving everyone in a company access to AI tools is that people will use them in unexpected ways. And that fact is steadily seeding fears that companies now have new, gaping security holes in their systems or, at least, will soon.