Rein Security raises $25 million to block harmful AI-agent actions as they happen

Rein Security has raised $25 million to fund a specific security promise: stop an AI agent’s harmful action without stopping its entire workflow. The October 8, 2026, Series A backs controls that operate while agents run. The company’s announcement describes customer adoption and rapid growth; those are company-reported signals, not independent measurements of how effectively its defenses work.

Glilot Capital and Sienna Venture Capital co-led the round, with Corner Ventures, Atlacle and RNP Capital Advisors participating. Rein says the financing brings its total funding to $35 million. It plans to spend the new capital on product development, research into AI-agent security and global hiring.

Intervening where an agent acts

Rein’s approach centers on runtime security—watching software while it executes, rather than only checking it beforehand. The company says its platform runs as a sidecar, a component alongside the application or agent. It monitors code execution and resource access without routing company or customer data through a gateway or proxy. Rein also says the platform secures the agent supply chain—the components agents rely on.

In an interview with SiliconANGLE, co-founder and CEO Matan Bar-Efrat described a system that establishes each agent’s normal behavior. It analyzes prompts, calls to other software and database access, then applies controls to individual actions that deviate from that baseline. His argument is that the action’s impact matters more than the attacker’s particular technique.

We aren’t terminating the agent process, thread or the request, we only prevent the specific malicious action from taking place while allowing the rest of the workflow to continue uninterrupted.

Matan Bar-Efrat, Rein Security co-founder and CEO, speaking to SiliconANGLE

Bar-Efrat frames the danger as downstream behavior: agents may act on instructions from other systems or agents, not just their human operators. Rein is positioning its platform both to secure agents a business deploys and to defend against AI-driven attacks. The common focus is preventing harmful execution, rather than predicting every possible way an attacker might manipulate an agent.

One example in the announcement involves an enterprise onboarding agent opening a seemingly ordinary PDF. Rein says hidden instructions—a prompt injection intended to redirect the agent—pushed it outside its assigned role. The company says it detected and blocked the action before damage occurred. It does not identify the enterprise or provide a detailed technical account of the intervention.

Rein also points to offensive research by its Agent Breakers team. At Black Hat USA 2026, the team presented work showing how it compromised an AI shopping agent at a top-five U.S. retailer, according to the company. SiliconANGLE dates the presentation to August and says the researchers also described how the vulnerability could be prevented. The retailer is not named.

Growth multiples without the starting scale

Rein says revenue has increased eightfold and its customer base fivefold since its January 2026 launch. The announcement gives neither starting nor ending revenue or customer counts. Those multiples describe a pace of growth, but leave the size of the underlying business unclear.

It names Flex and Swimlane as new customers, alongside Lemonade and Dun & Bradstreet, and says its platform secures thousands of agents executing millions of actions. That activity count describes the volume Rein says it protects; it is not a count of attacks prevented or a measure of detection accuracy.

The customer examples put those deployments closer to end users. Rein says it protects Lemonade agents supporting services used by more than three million active customers. At Dun & Bradstreet, it says the protected agents power products used by more than 240,000 customers. Those are the customers’ service audiences, not counts of Rein’s own paying customers.

Dun & Bradstreet’s chief cybersecurity and technology risk officer, Jay DePaul, supplies a customer perspective in the announcement. He says Rein helps the company see and control agent activity at scale. DePaul frames that visibility as an extension of the trust customers place in Dun & Bradstreet’s data as it puts agents into production.