Rubrik Launches AI Safety Net to Control Autonomous Agents

Rubrik introduced Agent Cloud and Rubrik AI last month to manage and mitigate risks associated with rogue autonomous AI agents [2, 3].

As enterprises deploy agentic AI to automate complex workflows, the potential for autonomous systems to make unauthorized or destructive changes increases. These tools aim to prevent systemic failures while allowing companies to pursue innovation and return on investment [1, 4].

Anneka Gupta, chief product officer of Rubrik, said the new capabilities were presented at the Rubrik Forward conference in Las Vegas [1, 2]. The platform provides a suite of controls designed to bring governance to autonomous systems, including auditability, and attribution [2, 3].

A central feature of the release is a safety net that functions as an "undo button," allowing administrators to reverse actions taken by an AI agent [2, 3]. This reversibility is intended to stop rogue AI from causing permanent data loss or operational disruption [2, 4].

Rubrik is positioning this infrastructure as a "billion-dollar safety net" ^[4] for the deployment of autonomous agents. By integrating these controls into its flagship cyber resilience platform, the company seeks to balance the speed of AI adoption with rigorous risk management ^[3].

The launch coincides with a broader industry shift toward agentic AI, where models do not just generate text but execute multi-step tasks independently. Rubrik's approach focuses on the recovery aspect of these interactions, ensuring that any autonomous error can be corrected without compromising the entire system [2, 3].