Washington, October 28, 2026 / PRNewswire / -- A security platform targeting proxy-based labor, Evoke Security, has today launched the industry's first assessment solution for identifying misuses of proxies and out-of-control proxies in major AI platforms and coding toolchains. This solution, named "Out-of-Control Proxy Assessment" (Rogue Agent Assessment), can help organizations understand how employees use proxies, when proxies are misused, and when proxies violate explicit instructions.
As employees integrate the use of proxies into their daily work processes, security teams are facing a challenge that existing security tools have not yet addressed. Proxies may take actions that expose sensitive data, expand access privileges, or create unnecessary risks while completing assigned tasks. Nowadays, it is no longer sufficient to simply know whether a proxy has followed instructions to determine whether its operation is secure.
Evoke Security, co-founder and CEO, Jason Rebholz stated: "Every organization that collaborates with us is facing the same two types of risks: employees misusing proxies, and proxies getting out of control. Proxies are bypassing security controls and taking actions that no security team would approve. An individual proxy action may seem safe on its own, but it is only by observing a series of proxy actions that we can get a complete picture of the situation. This is why Evoke monitors the entire trajectory of proxies, identifies when they are heading towards high-risk paths, and guides them back to safe paths or stops them outright before any damage is caused."
Evoke indicates that the uniqueness of this industry's first assessment solution lies in its focus on monitoring the current and historical behavior of proxies, rather than just misconfigurations. Relying on the experience of its founding team in incident response and the Chief Information Security Officer ( CISO ), Evoke has developed a framework to identify out-of-control proxies from four dimensions.
- Task ( Mission ): Is this action consistent with the initial goal of the agent?
- Method ( Method ): Is this action reasonable for the task?
- Motivation ( Motive ): What intention does this behavior imply?
- Impact Scope ( Magnitude ): How does this action expand the radius of potential risk impact?
The company stated that this is one layer of the Evoke in-depth detection method, which helps security teams distinguish between proxies that are operating as expected and those that introduce substantial risks. This framework does not consider every abnormal action as a threat; rather, it helps to determine whether a proxy should continue to execute, be directed to a safer path, or should be stopped.
Evoke indicates that the company is actively cooperating with native AI enterprises to monitor and control AI proxy behaviors. Assessments show that proxies can also introduce new risks when performing legitimate tasks, which further highlights the importance of real-time monitoring and guiding proxy behavior. In one assessment, Evoke discovered that a coding proxy performing security reviews escaped from its own sandbox and later even lied to users.
"Out-of-control proxy assessment" provides an organized way for organizations to evaluate their proxy security posture from three key areas.
- Misuse of proxies: Identify instances where proxies are used to bypass existing security controls or violate usage policies.
- Out of control proxy: Identify situations where proxies stop following instructions or take unnecessary, high-risk paths.
- Shadow AI: Unapproved use of AI, MCP, or Agent Skills products was detected in the environment.
The company stated that by combining these capabilities, Evoke provides organizations with an assessment specifically designed to address the security challenges of autonomous proxies. Security teams will have a clearer understanding of which proxies exist, how they operate, where the risks lie, and what actions are required for intervention.
Evoke indicates that the "Out-of-Control Proxy Assessment" is now available to organizations that wish to gain a better understanding of the exposure level of their AI proxies and to establish effective security boundaries without imposing excessive restrictions on their operations.
About Evoke Security
Evoke Security is a AI detection and response platform specifically designed to protect AI proxies. The company was founded by seasoned professionals in the field of incident response, and its platform helps security teams to continuously inventory, establish threat models, and monitor AI proxies. Evoke views proxies as the next generation of enterprise operating systems, implementing strict controls by identifying high-risk configurations and analyzing real-time behaviors and actions, ensuring that enterprises can advance innovation without accumulating security liabilities.
