Data Security , AI/ML , Governance, Risk and Compliance , RSAC Trust or fail: AI unlocks the value of unstructured data but raises new challenges for AI success April 28, 2026 Share By Paul Wagenseil Credit: Adobe Stock Images Agentic AI will transform work because it can automate decisions and actions at machine speed. But this comes with a fundamental risk because AI is only as trustworthy as the data it uses. Today, most enterprise data — especially unstructured data — is poorly understood, loosely governed, and hard to secure. Because of this, scaling AI safely requires a unified way to map, understand, and control data itself, a concept that Veeam calls a "Data Command Graph." Why AI success depends on trusted, governed, and recoverable data In-house AI systems depend on vast amounts of data drawn from across the enterprise, including structured databases, unstructured files, emails, documents, and data generated or modified by AI agents themselves. But without a clear understanding of where this data is stored, how it is classified, and who or what can access it, organizations cannot ensure reliable AI outcomes. One solution to this challenge is to create a unified map of all data and its relationships, showing how data connects to identities, permissions, applications, and AI agents. The map provides context essential for identifying sensitive data exposure, misconfigured access controls, unauthorized AI activity or other risks. This is the Data Command Graph, embedded at the center of Veeam's Agent Commander platform. "Simply connect Agent Commander to your cloud, your SaaS, or your AI systems," says Rehan Jalil , President of Security and AI at Veeam. "In minutes, you will see AI agents and AI models in your environments, how they are truly related to users and human identities, the data they can access, and what is sensitive inside that data." The graph also makes continuous governance possible. As AI agents interact with data, the graph tracks those interactions in real time, making sure that policies are enforced and that data usage remains compliant. Why data resilience and AI trust are board-level priorities Because the consequences of AI failure — hallucinated outputs, data leaks, and compliance violations — often have direct financial and reputational impacts, data resilience and AI trust are now board-level concerns. Organizations must show that they not only can protect data but recover from errors quickly and precisely. Traditional recovery strategies, such as broad system rollbacks, may be too disruptive or imprecise for AI-driven environments. A Data Command Graph solves this problem by providing end-to-end visibility and control. It reveals how data flows across systems, how permissions have been applied, and how AI agents interact with sensitive information. This lets organizations spot hidden risks, such as shadow AI usage or toxic combinations of permissions, before they can lead to incidents. How unified platforms with real-time visibility and precise recovery enable safe AI at scale To safely scale AI, organizations need the ability to act upon detecting a threat. Unified platforms built around a Data Command Graph provide this capability by combining detection, protection, and remediation. Real-time visibility lets organizations see how data, identities, and AI agents interact. When anomalies arise — such as an AI agent accessing sensitive data or performing unauthorized actions — policy-driven controls can intervene right away, stopping risks from escalating and ensuring that AI operates within defined boundaries. Equally important is the ability to undo mistakes. Agentic errors can propagate rapidly and affect multiple systems and datasets. The Data Command Graph enables what Veeam describes as "surgical recovery," or the ability to reverse specific AI-driven actions without disrupting entire workloads, providing precision critical for maintaining operational continuity while correcting issues. "You can also see if your data is resilient against AI mistakes in a backup system," explains Jalil, "ready to be recovered if AI actually misbehaves or becomes malicious. Ultimately, this unified approach bridges the gap between innovation and control. It lets organizations proceed with AI adoption while retaining confidence in their data, systems, and outcomes. An In-Depth Guide to AI Get essential knowledge and practical strategies to use AI to better your security program. Learn More Paul Wagenseil Paul Wagenseil is a custom content strategist for CyberRisk Alliance, leading creation of content developed from CRA research and aligned to the most critical topics of interest for the cybersecurity community. He previously held editor roles focused on the security market at Tom’s Guide, Laptop Magazine, TechNewsDaily.com and SecurityNewsDaily.com. Related Data Security Malicious elementary-data package version 0.23.3 steals developer data and cryptocurrency wallets SC Staff April 28, 2026 The attack exploited a GitHub Actions script injection flaw, allowing the attacker to inject shell code that exposed a GitHub token. Data Security Udemy allegedly breached by ShinyHunters, data leak warned SC Staff April 28, 2026 Cybernews reports that U.S-based. e-learning platform Udemy was claimed to have more than 1.4 million records compromised by the ShinyHunters extortion group, which threatened to expose the data trove should the firm opt out of negotiations by Apr. 27. Security Operations French police arrest hacker ‘HexDex’ for alleged widespread data theft SC Staff April 27, 2026 The investigation began in late December 2025 following approximately 100 reports of data theft. Related Events Cybercast Beyond the Hype: The Cybersecurity Trends CISOs are Keeping an Eye on in 2026 On-Demand Event Cybercast Beyond the data perimeter: Why next-generation DSPM is the foundation for modern data security On-Demand Event Virtual Conference Securing the Future of Finance: Strategies to Counter Modern Cyber Threats On-Demand Event Get daily email updates SC Media's daily must-read of the most current and pressing daily news Business Email By clicking the Subscribe button below, you agree to SC Media Terms of Use and Privacy Policy . Subscribe Related Terms British Standard 7799 Business Impact Analysis (BIA) Checksum Cipher Ciphertext Competitive Intelligence Data Encryption Standard (DES) Data Warehousing Digital Envelope Digital Signature Standard (DSS) You can skip this ad in 5 seconds