Agentic AI Attacks Force Security Back to Data Source

Agentic AI is reshaping cybersecurity by pushing security controls directly to the database level, where access policies can be enforced regardless of what instructions an AI agent receives. At Oracle's AI cyberattack event, theCUBE Research's Krista Case and Dave Vellante discussed how agentic-fueled attacks have exposed limits in even mature security programs. They outlined an emerging shared accountability model where vendors ensure safe, governed product behavior while customers assess asset criticality and risk tolerance. Recovery now extends beyond restoring data to restoring trusted business processes and functions disrupted by AI-driven incidents.