Commvault Revolutionizes Cloud with an 'Undo' for AI

Commvault Introduces an Innovative Cloud Solution

Companies using cloud environments can now benefit from a rollback feature for artificial intelligence agents, thanks to the launch of Commvault AI Protect. This new tool aims to provide enhanced security by allowing users to revert actions taken by autonomous models operating in infrastructures such as AWS, Microsoft Azure, and Google Cloud.

As autonomous software evolves, these systems are capable of managing complex tasks such as deleting files, accessing databases, creating server clusters, and even rewriting access policies. Commvault has identified the governance risks associated with these capabilities and developed AI Protect to monitor and control these agents.

The Challenges of Traditional Governance in the Face of AI

Traditional governance relies on static rules, where a human user is granted specific permissions to perform predictable tasks. In case of an issue, responsibility is clearly defined. However, AI agents exhibit emergent behaviors that complicate this approach.

When an AI agent receives a complex request, it may unexpectedly combine approved permissions to solve the problem. For example, an agent might decide to delete a production database to optimize storage costs, executing this action in mere milliseconds.

Unlike a human engineer who might hesitate before executing a destructive command, an AI agent simply follows its internal logic. It is capable of executing thousands of API requests per second, making it difficult for human security operations centers to respond in time.

A Tailored Response for Agentic Environments

Pranay Ahlawat, Chief Technology and AI Officer at Commvault, explains that in environments where agents rapidly alter the state of data and systems, it is crucial to be able to recover not only the data but also the entire technology stack to revert to a stable state.

AI Protect represents a new generation of governance tools for AI agents in the cloud. It continuously scans the company's cloud footprint to identify active agents, an essential task in the face of hidden AI that poses significant challenges to IT departments.

Detection and Monitoring of Hidden AI Agents

Developers often launch experimental agents with corporate credentials without informing security teams, connecting language models to internal data lakes to test new workflows. AI Protect forces these hidden agents to be identified and monitored.

Once detected, the software monitors specific API calls from the agent and data interactions on AWS, Azure, and GCP. Every database read, every storage modification, and every configuration change is recorded.

A Safety Net for Cloud Environments

The rollback feature of AI Protect provides an essential safety net. If an AI model hallucinates or misinterprets a command, administrators can restore the exact state of the environment before the machine initiates a destructive sequence.

However, the cloud infrastructure is highly stateful and interconnected. Reversing a complex chain of automated actions requires precise tracking based on a ledger. It is not enough to restore a single database table if the machine has also modified networking rules or altered access management policies.

Commvault's Approach to Enhanced Security

Commvault combines traditional backup architecture with continuous cloud monitoring to achieve this goal. By mapping the blast radius of the agent's session, the software isolates the damage and untangles the specific changes made by the AI from legitimate modifications performed by human users.

This prevents a massive rollback from inadvertently deleting valid customer transactions or erasing hours of legitimate engineering work. Machines will continue to execute tasks faster than human operators can monitor them, but the goal now is to implement protective measures that ensure autonomous actions can be instantly and accurately undone.