Agentic AI puts new pressure on identity and database security
SiliconANGLE Mark Albertson ● Covered by 3 sources
Agentic AI is pushing companies to rethink identity and database security. Oracle says the big task is knowing where sensitive data lives before AI agents start using it.
Based on reporting by SiliconANGLE, Mark Albertson — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Security teams are shifting fast as agentic AI moves into enterprise work. The new job is not just spotting threats, but automating enough to turn signals into action without losing context or control. That sounds tidy on paper. In practice, it means figuring out where AI should act on its own and where humans still need to hold the wheel.
Oracle is trying to meet that shift from the database side. At the company’s “AI Cyberattacks Are Escalating: How to Secure Your Data Now” event, Johnnie Konstantas, group vice president for Oracle Cloud Engineering, said customers want automation, but only after they understand where sensitive assets sit and where AI is being built inside the company. He called that inventory extremely important. No mystery there: if you do not know what you have, you cannot protect it.
The bigger security headache is identity. Konstantas said agentic AI changes the meaning of permissions because an agent, and even several sub-agents, can act as a proxy for a person and carry those permissions into a workflow. That raises the stakes for keeping identity uniform and pervasive. It is a pretty direct warning: once AI can act, identity stops being a log-in problem and becomes an access-and-action problem.
Oracle says it has folded agentic AI into its security portfolio, and Konstantas argued the best answer starts with the customer’s data, not with Oracle’s product pitch. He pointed to capabilities in the latest version of its AI database that are aimed at agentic AI. The underlying message is simple enough. AI is moving into the parts of the enterprise that matter most, so security has to move there first too.
My take — AI-written commentary, not fact-checked reporting
This is the part of AI everyone likes to skip: autonomy is adorable until it starts wearing your permissions badge. Oracle is pushing the right message here, even if the sales pitch is doing some heavy lifting in the background. Identity and data controls are the real bill coming due for agentic AI, and the companies pretending otherwise are basically inviting a mess.
Read more about this at: SiliconANGLE