Data security company Rubrik has officially launched Agent Cloud, a security service dedicated to Anthropic's Claude Code and Claude Cowork, SiliconANGLE reported on June 9 (local time).

Rubrik unveiled a range of new products for AI agent security and recovery at the Rubrik Forward conference in Las Vegas.

Rubrik co-founder and CEO Bipul Sinha (비풀 신하) said AI agents can be abused by external attacks or cause problems during internal deployment. He said Rubrik AI provides resilience against machine-speed cyber intrusions and compromised agents.

The company said Agent Cloud for Claude has four core functions. The semantic AI governance engine SAGE controls agent behavior in real time based on intent rather than static rules. Agent Inventory identifies the risk level, access privileges and policy violations of deployed agents.

Agent Rewind rolls back all unintended agent actions, including those involving Claude Code and Cowork. Codebase Resilience stores permanent snapshots of GitHub and Azure DevOps repositories externally, enabling recovery even if an agent or attacker overwrites commit history or deletes branches.

Rubrik Chief Product Officer Anneka Gupta (아네카 굽타) said organizations are adopting Claude faster than any other agentic technology. She said it provides what agents can access, what they do, rollback of actions, and recovery functions for code and agent configurations.

Through Project Hourglass, Rubrik also signed partnerships with six global systems integration firms: Cognizant, Deloitte, HCLTech, NTT Data, Wipro and LTM.

According to a survey by Rubrik's in-house research lab, 86 percent of cybersecurity leaders expected AI agents to bypass security guardrails within 1 year. Another 88 percent are concerned that rising agentic threats will make it harder to meet recovery time objectives.