AI agents expose critical gaps in corporate data protection

October 19, 2026 Veeam 4 min read

More than 82% of companies have unknown AI agents operating in their IT environments, while 95% say data-related challenges have already slowed their AI progress.

Table of Contents

Although 88% of organizations already use or are testing AI agents, only 7% are truly prepared to adopt them, according to a Veeam report.

The use of artificial intelligence agents in businesses poses new challenges for data protection, particularly when these systems can automatically access, modify, or delete corporate information. Visibility into their activities and the ability to recover specific information are becoming important aspects of organizational resilience.

In this context, Veeam Software announced the general availability of Intelligent ResOps for Microsoft 365, a solution that enables organizations to identify AI agents interacting with data, detect file modifications, locate sensitive information, and precisely recover affected items. The product complements Data Cloud for Microsoft 365 and can be integrated without migrating data or redesigning the existing architecture.

The company also announced private previews of Intelligent ResOps for Data Platform and Okta, aiming to extend monitoring and recovery capabilities to infrastructure and identity management environments. For the fourth quarter of 2026, it plans to begin private testing of solutions for GitHub, Azure DevOps, and Google IAM.

The 2026 Veeam Data Trust and Resilience Report states that more than 82% of companies have unknown AI agents operating in their IT environments and that 95% report that data-related challenges have already slowed their progress in artificial intelligence. The findings reflect the difficulties organizations face in monitoring automated actions and maintaining control over their information.

Related Posts:

Monitoring AI agents in Microsoft 365.

The Microsoft 365 solution incorporates DataAI Command Graph, which connects production data with backup context. Through this integration, teams can identify agents such as Copilot, review what sensitive information they can access, and detect changes made to files, with the option to reverse specific modifications.

It also enables the classification of sensitive and regulated data in SharePoint, OneDrive, Teams, and Exchange, as well as the identification of redundant, obsolete, and trivial (ROT) information for quarantine while determining whether it should be archived or deleted. These capabilities aim to reduce storage costs and facilitate the identification of access and protection risks.

Another capability is the verification of backup coverage for business-critical data and the detection of gaps before an incident occurs. Authorized team members can also ask natural-language questions about data, permissions, AI agents, and backup protection through the integrated agents.

Anand Eswaran, CEO of Veeam: “For twenty years, the question of resilience was simple: can I recover? In the era of agentic AI, the question is whether you can precisely determine what has been affected, who or what changed it, and how to return to an exact trusted state. Intelligent ResOps gives customers both a recovery point and a recovery map. A recovery point tells you what can be restored. A recovery map tells you what should be restored first, what has changed, and what should not be recovered. This is only possible when advanced contextual intelligence is connected to data, identities, access, and activity across the entire data environment. With iResOps for Microsoft 365, we combine the market-leading resilience foundation already trusted by more than 26 million users with Securiti AI’s advanced capabilities in data security, governance, and AI to help customers adopt AI with greater confidence.”

Expansion into infrastructure and identity systems.

The private preview for Data Platform connects existing backup servers to DataAI Command Platform without requiring migrations or major architectural changes. The console brings together information about protected infrastructure, workloads, and indexed files, and enables users to search production and protected data, detect under-resourced servers, and recover files or infrastructure with the context of their restore points.

This version includes role-based access controls, single sign-on, and multifactor authentication, as well as three integrated agents: Dashboard Agent, for querying information; Graph Agent, for exploring data; and Backup Admin Agent, for improving operational efficiency. General availability is planned for the end of the fourth quarter of 2026, alongside Veeam Data Platform v13.2.

In the identity space, Intelligent ResOps for Okta creates immutable restore points for users, groups, applications, policies, and administrator roles, covering 29 object types. The solution allows organizations to compare the live environment with a trusted state, identify changes, and restore objects along with their relationships, such as group memberships and application assignments.

Recovery also covers elements such as federation and roles, with the aim of correcting specific changes without manually rebuilding access permissions or rolling back the entire environment. According to the company, each object’s restorability is verified through Okta’s live API. The service extends the identity resilience capabilities the company already offers for Active Directory and Microsoft Entra ID.

Availability and upcoming developments.

The Microsoft 365 solution is already available as an add-on to Data Cloud for Microsoft 365, in the Foundation and Advanced editions, and through Veeam Cloud & Service Provider (VCSP) partners.

The Data Platform and Okta versions are available in private preview to eligible customers.

The company plans to expand private testing to GitHub and Azure DevOps during the fourth quarter of 2026, with protection and recovery capabilities for code, workflows, configuration, and permissions. It also plans a preview of Intelligent ResOps for Google IAM for selected customers.

With this expansion, the company is extending its data resilience offering to environments where AI agents, applications, and identity systems interact with corporate information, focusing on monitoring changes and selectively recovering affected items.