Pluto Security Raises $42M to Tame Shadow AI Agents

Corporate technology teams are losing track of the software running on their company laptops. For two decades, chief information security officers maintained iron control over corporate networks through strict device management, endpoint agents, and locked administrative permissions. An employee who wanted to install a new program had to submit an IT ticket, wait two weeks for security vetting, and receive formal approval. That defensive perimeter dissolved almost overnight as machine assistants entered web browsers. Workers now paste proprietary customer records into unvetted browser tabs, link autonomous browser extensions to corporate email accounts, and build automated desktop workflows with a few plain sentences. Israeli cybersecurity venture Pluto Security stepped out of stealth on Monday to tackle that blind spot. The firm announced $42M in total funding, combining an $8.5M seed round led by TLV Partners with a $34M Series A led by Jump Capital. Skinos, Ansa Capital, and the founders of security unicorn Armis joined the round. The capital arrives as automated tools create severe perimeter risks for IT managers, an issue we documented when Cloudflare deployed defensive software to harden firewalls.

The financing highlights rapid commercial demand from businesses worried about intellectual property leaks. Founded by chief executive officer Shahar Bahat alongside technical leaders Tal Ofer Regev and Gil Maman, Pluto Security built an inspection platform that tracks how corporate employees interact with machine learning tools, browser add-ons, and autonomous scripts. The company passed $2M in annual recurring revenue within months of initiating commercial sales, expanding to more than fifty workers across offices in Tel Aviv and New York. Security leaders are purchasing the platform because legacy monitoring tools cannot see inside modern web sessions. When an employee instructs a browser extension to summarize an internal document, traditional network firewalls see only encrypted HTTPS traffic moving to a reputable cloud host. Pluto monitors the local execution layer, catching prompt injection attempts, sensitive data transfers, and risky agent actions as they happen. The rush to secure workplace tools matches investment trends we followed when Armadin raised $255M for agent swarm security.

The Hidden Perimeter of Employee Built Software

To understand why venture investors placed $42M into an early-stage security firm, one must look at how corporate software development changed over the past eighteen months. Writing custom business tools once required trained computer engineers working inside controlled code repositories. Today, ordinary office workers in marketing, sales, and accounting use conversational coding assistants to build custom scripts that move customer records between databases.

This shadow automation creates immense vulnerability. A sales manager with zero security training can configure an automated agent to scrape customer contact lists, connect the output to an outside messaging platform, and execute outreach campaigns without notifying the IT department. If that homemade script contains an unpatched security flaw or uses weak authentication keys, attackers can hijack the connection to extract customer databases. Bahat, who previously held senior engineering roles at Microsoft and Palo Alto Networks, noted that enterprise security teams are not trying to ban machine tools; they simply need to know what scripts are doing with company data. Pluto software maps these ad-hoc tools, giving IT directors complete visibility over employee automations. How corporate platforms handle internal software permissions was detailed when we covered Anthropic tightening network defenses after autonomous software incidents.

Tracking Execution at the Endpoint Browser Interface

The technical architecture behind Pluto platform avoids the latency penalties of traditional web proxy servers. Older enterprise security tools route all corporate internet traffic through centralized inspection gateways. That approach creates severe network delays, degrades video conference connections, and breaks when employees travel outside corporate virtual private networks.

Pluto operates directly at the local device layer, monitoring browser processes and operating system calls through a lightweight endpoint sensor. When an employee opens a conversational assistant or activates an autonomous browser tool, the software inspects the text input, the destination API endpoint, and the requested file permissions locally. If a financial analyst attempts to upload an unredacted spreadsheet containing Social Security numbers or credit card balances, the system intervenes instantly. It can redact sensitive columns automatically, block the transfer, or alert the compliance team without halting the employee broader workflow. This granular policy enforcement prevents data leaks while allowing staff to use modern software assistants safely. The challenge of securing corporate data against unauthorized model access was explored in our analysis of investigations into models accessing federal records.

Defending Against Prompt Injection and Hijacked Agents

Beyond tracking outbound data leaks, Pluto protects corporate workers from incoming manipulation attacks. As enterprises connect autonomous agents to email accounts and customer service ticketing queues, hackers have developed indirect prompt injection tactics. An attacker hides malicious instructions inside an ordinary customer support inquiry or public web page, designed to hijack the agent when it reads the text.

If an automated customer service agent reads an incoming email containing hidden commands, the rogue text can instruct the agent to forward internal files or reset customer passwords. Traditional antivirus software cannot spot these attacks because the malicious payload consists of plain English sentences rather than binary exploit code. Pluto inspects conversational streams for adversarial instruction patterns, quarantining compromised prompts before the host agent can execute unauthorized system actions. Regev, who previously managed more than 150 engineers as vice president of research and development at Palo Alto Networks, designed the inspection engine to identify behavioral manipulation in real time. The rising operational hazards of automated cyberattacks were examined in our review of machine tools lowering technical barriers for digital intrusions.

Founding Heritage From Unit 8200 and Tech Titans

The credibility driving Pluto early sales momentum stems from its technical founding pedigree. Maman spent more than six years in Israeli military intelligence unit 8200, leading advanced cyber research teams that analyzed nation-state intrusion tactics. That defense experience, paired with Bahat and Regev enterprise software backgrounds, gave the startup immediate access to chief information security officers at major American corporations.

Early-stage enterprise sales often stumble because young startups cannot satisfy strict corporate compliance audits. Pluto team used its industry connections to land paid production deployments with Fortune 500 financial institutions, healthcare providers, and technology firms within months of writing its first line of code. Investors in the round noted that seeing a security startup achieve multi-million ARR so quickly is rare in a cautious venture market. The capital will fund research laboratories in Tel Aviv and support sales expansion across North America. How elite engineering teams turn technical expertise into enterprise software sales was analyzed when venture investors detailed founder testing before committing capital.

The Regulatory Squeeze on Corporate Data Governance

The rush to install AI monitoring platforms is also being propelled by aggressive regulatory enforcement. Government watchdogs in North America and Europe are penalizing companies that fail to control how automated tools process personal consumer information. Under the European Union AI Act and revised Federal Trade Commission guidelines, corporate boards are legally responsible if employee-built models mishandle customer data or generate discriminatory outputs.

Corporate risk officers can no longer plead ignorance when an employee uses an unauthorized web tool that exposes customer files. Regulatory agencies demand verified audit trails showing that corporate data is protected from third-party model training loops. Pluto automatically logs every interaction, creating tamper-proof compliance records that demonstrate corporate governance during regulatory examinations. Providing verifiable proof protects enterprises from costly statutory penalties, a dynamic we followed when Helmguard raised capital to automate governance workflows.

The Venture War for AI Security Dominance

Pluto successful Series A arrives amid intense competition across the cybersecurity sector. Established security giants like CrowdStrike, Palo Alto Networks, and Zscaler are racing to add generative defense modules to their existing platforms. At the same time, venture capital funds are backing dozens of early-stage startups promising to secure enterprise machine learning deployments.

To maintain its growth trajectory, Pluto must prove that its specialized focus on browser tools and autonomous desktop agents provides deeper protection than bundled features from legacy security vendors. Saurabh Sharma, partner at Jump Capital, argued that securing autonomous agents requires a ground-up architectural rebuild that older endpoint protection platforms cannot deliver through simple software patches. Pluto bet is that as agents take over daily office work, specialized agent security will become as mandatory as firewalls were during the dawn of the public internet. The corporate competition shaping specialized security tools matches trends we tracked when CrowdStrike partnered with OpenAI to protect autonomous systems.

The Balance Between Employee Speed and Corporate Safety

The true operational victory of Pluto platform lies in its approach to employee productivity. In earlier computing eras, corporate security teams acted as the Department of No, blocking new technologies, restricting website access, and forcing staff to use outdated internal software. That rigid defensive posture no longer works; when IT departments block useful tools, employees simply route around the restrictions using personal smartphones and unmonitored home connections.

By providing real-time guardrails rather than blunt bans, Pluto allows enterprises to embrace modern workplace tools without compromising digital perimeters. Workers can use conversational assistants to summarize long research files, draft software code, and automate repetitive office tasks, confident that the security software is protecting corporate secrets in the background. As autonomous software agents become standard digital coworkers, the companies that thrive will be those that learn to govern their digital workforce safely. Pluto $42M funding proves that the future of enterprise defense is not about stopping technological progress, but building the safety fences that make progress possible.

Read More on TechRobust:

• Cloudflare uses agent harness to defend web firewalls
• Armadin secures $255M Series B for agent swarm security
• CrowdStrike and OpenAI expand partnership to secure AI agents