The agentic AI era is forcing security and resilience into a single discipline

The agentic AI era is forcing security and resilience to merge into a single discipline. Traditionally, security focused on keeping bad actors out, while resilience helped businesses recover after being breached. However, agentic artificial intelligence is blurring these lines, and most organizations have yet to notice. Veeam Software Group GmbH acquired Securiti AI, positioning itself as the first AI resiliency company.

At VeeamON 2026, they introduced Veeam Intelligent ResOps (iResOps), initially available for Microsoft 365. Whether referred to as AI resiliency or resilience operations, the concept is the same: when software agents act on data at machine speed, one can't secure what they can't recover, and one can't recover what they don't understand.

During a keynote at VeeamON Global Launch, Veeam's CEO Anand Eswaran illustrated the issue with a Spanish company that had an AI agent breach its systems by changing personal data and billing records. The key problem lies in disclosing modifications, as it's impossible to reverse what one doesn't precisely identify. Eswaran distinguished three eras: backup and recovery, cyber resilience, and the agentic era.

All three now coexist. Agents are often temporary, but data and its drift are not. A well-intentioned agent can still cause damage by reasoning over incorrect data. Most of the industry's approach to agent risk focuses on the agent itself, but Eswaran argued that understanding data is more crucial, as every agent eventually touches it.

Another critical point Eswaran made was the difference between recovery point and recovery map. While a recovery point tells you what you can restore, a recovery map tells you what to restore first, what changed, and what not to bring back. This distinction is essential as recovery often becomes an all-or-nothing process, leading to increased costs and time.

Veeam's iResOps for Microsoft 365 aims to provide a unified solution, discovering AI agents, showing sensitive data access, flagging redundant or obsolete data, and enabling teams to undo changes. Private previews in data science environments like Veeam Data Platform, Okta, GitHub, Azure DevOps, and Google IAM are planned.