Airlock Digital blocks unauthorized AI agent actions at the endpoint
Airlock Digital has launched Agentic Control to manage AI agent actions at endpoints, addressing security concerns as AI adoption grows.
Airlock Digital has announced the release of Agentic Control with Airlock Digital v7.1, a feature designed to evaluate and manage the actions of AI agents at endpoints, thereby enhancing security protocols for organizations adopting this technology.
Airlock Digital has unveiled a new feature called Agentic Control, included in the recently released Airlock Digital v7.1. This feature allows organizations to extend their existing Deny by Default Application Control to encompass AI agents.
Traditionally, application control focused on managing what software employees could run on their devices. However, AI agents now operate on the same endpoints, often with the same access as their human counterparts. These agents can autonomously execute commands, access files, and adapt their actions to achieve specific goals. This capability raises important questions for security teams about the permissions and actions that should be allowed for these AI agents.
Agentic Control is designed to tackle this uncertainty by assessing an AI agent's actions against the organization's policies before execution. If an action is deemed unauthorized, the system prevents it from occurring and provides feedback to the agent, allowing it to adjust its behavior while remaining compliant with policies. This proactive approach aims to mitigate the risks associated with AI agents taking potentially harmful actions in pursuit of their objectives.
The rise of AI agents has led to significant security challenges. A report from the Cloud Security Alliance in April 2026 indicated that 82% of organizations had unidentified AI agents operating within their environments, and 65% had faced security incidents related to these agents in the previous year. The introduction of Agentic Control, alongside Airlock Digital's AI discovery features, equips security and IT teams with enhanced capabilities to manage these emerging threats effectively.
Agentic Control introduces a new layer of preventative checks at the endpoint, where AI agent actions are initiated. This feature complements existing security measures such as identity and access management, data security, and detection and response systems. By determining whether each action aligns with organizational policies, Airlock Digital helps organizations maintain control over AI agent behavior.
Company executives have highlighted the urgent need for such solutions. Kevin Dunne, CEO of Airlock Digital, noted that the pace of AI agent adoption is outstripping the security measures designed to govern their use. Agentic Steering, a component of Agentic Control, is intended to provide the visibility and control necessary for organizations to embrace AI confidently without hindering business operations.
While Airlock Digital's Agentic Control offers a promising solution for managing AI agents, several unknowns remain. It is unclear how organizations will implement these controls across diverse environments and what the overall effectiveness will be in real-world applications. Additionally, the long-term impact of these measures on AI agent behavior and security incident rates will require ongoing observation.
What it meansAirlock Digital has launched Agentic Control to manage actions of AI agents at endpoints, addressing security issues as AI usage increases.
Ask your IT team this weekWhat policies do we currently have in place for AI agent actions?How are we monitoring the behavior of AI agents in our environment?
DecisionThe IT team will implement policies for AI agent actions, and security will verify compliance.