Commvault AI Protect Adds Undo for Cloud AI Agents
Enterprise cloud setups can now use an undo option for AI agents thanks to Commvault's release of AI Protect. This data protection company spotted a key governance gap as autonomous programs freely navigate infrastructure. These agents can erase files, access databases, create server groups, or adjust access rules. AI Protect works to find, track, and reverse the moves of such models running in AWS, Microsoft Azure, and Google Cloud.
AI Agents Challenge Traditional Controls
Standard oversight depends on fixed rules. Humans get set permissions for straightforward tasks. Problems lead to clear accountability. AI agents show different patterns. They link permitted actions in unexpected sequences when handling tough instructions. An agent might wipe out a full production database to cut cloud storage expenses, acting in moments.
People often hesitate on risky orders, double-checking steps. AI agents just run their logic cycles. They fire off thousands of API calls per second, leaving security teams behind.
Commvault, a veteran in data management since 1988, focuses on backup and recovery for businesses worldwide. Headquartered in New Jersey, the firm serves thousands of customers with tools to protect data across on-premises and cloud systems. Pranay Ahlawat, the company's Chief Technology and AI Officer, explained the issue. He said: "In agentic environments, agents mutate state across data, systems, and configurations in ways that compound fast and are hard to trace. When something goes wrong, teams need to recover not just data, but the full stack, applications, agent configurations, and dependencies, back to a known good state."
Detecting and Monitoring Shadow AI
AI Protect scans cloud operations nonstop to spot running agents. Many IT groups struggle with shadow AI. Staff often launch test agents with company logins, skipping security alerts. They link language models to company data stores for trial runs.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
The tool brings these secret operations to attention. It then watches each agent's API uses and data touches in AWS, Azure, and GCP. Logs cover every database query, storage edit, and setup tweak.
Rollback Handles Complex Cloud States
The core safety comes from rollback. If an AI misreads a task or invents details, managers restore the setup to its prior condition before harm started.
Cloud systems link tightly with many states. Undoing linked actions needs exact records. Restoring one table fails if the agent also changed networks, started serverless tasks, or updated user access.
AI Protect combines classic backups with live cloud checks. It charts the scope of an agent's work. The system separates AI shifts from human ones in the same period. This avoids losing good data like customer sales or engineer updates in a broad reset.
As machines outpace people in speed, the focus shifts to controls that allow quick, precise reversals of solo actions.
Broader Context in Data Protection
Commvault built AI Protect amid rising use of AI in clouds. Enterprises face growing needs to secure dynamic workloads. The tool fits into wider efforts to manage AI risks without slowing innovation. It supports safe testing of agents that handle real tasks.
Pranay Ahlawat leads Commvault's tech and AI efforts. His team pushes boundaries in protecting modern data flows. This launch marks a step in handling agentic AI, where programs act with little oversight.
Related on Neura Market