The Point: Versa Verbo implements policy-based approval workflows for AI agents with complete verification of each action based on identity, roles, and policies.
SASE provider Versa has introduced a zero-trust architecture for the Model Context Protocol (MCP) to verify AI agent actions before their execution. The solution prevents uncontrolled or misinterpreted processes from compromising critical systems.
As automation through AI agents increases, new control challenges emerge: a simple prompt can trigger a multitude of processes that affect networks, security policies, or critical systems. The risk lies in AI systems misinterpreting intentions or executing unintended actions – without prior approval.
Versa’s approach is based on the zero-trust principle: no action triggered by an AI is automatically classified as trustworthy. Instead, the system verifies each step based on user identity, role permissions, and predefined policies. For critical operations, manual approval by humans may additionally be required. Administrators define in advance which tasks AI agents may execute independently, which require approval, and which are blocked – differentiated by user role, system context, and risk level.
The solution is integrated into Versa Verbo, an AI assistant of the VersaONE Universal SASE Platform. This combination enables event correlation, anomaly detection, and guided troubleshooting. All approved steps are logged, so that organizations can later trace which decisions the AI made. This is intended to make AI-supported processes usable even in production environments without losing control.
The zero-trust MCP architecture is available as of version 23.1.1 of the VersaONE Universal SASE Platform. Versa Verbo supports cloud, on-premises, and hybrid deployments, allowing organizations to decide for themselves which models to use and where data is processed.
Source: www.it-daily.net · Published June 26, 2026
Lumi AI News — AI-assisted curation in accordance with Art. 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.1.