What to expect during Oracle’s ‘AI Cyberattacks Are Escalating: How to Secure Your Data Now’ event: Join theCUBE on Sept. 22

Oracle is leaning hard into a simple idea: if AI is changing the threat, security has to move closer to the data itself. That is the center of its “AI Cyberattacks Are Escalating: How to Secure Your Data Now” virtual event on Sept. 22, where Oracle leaders and theCUBE Research analysts will talk through what breaks, what still works and what needs to change.

The company’s answer is a three-part strategy: secure at source, secure at speed and secure through resilience. In practice, that means pushing controls into the data layer, where Oracle says policy can be enforced consistently even as AI agents start acting on behalf of users across more workflows.

Oracle has already started building that out with tools such as Oracle Deep Data Security and Oracle SQL Firewall. Deep Data Security applies end-user-specific privacy rules inside the database, while SQL Firewall is aimed at SQL injection attacks, where malicious commands are slipped through website input fields. Oracle also said in April that its Oracle AI Database got enhancements for centralized, declarative, fine-grained authorization and data visibility policies tied to identity, role and context.

The bigger point here is governance. theCUBE Research’s Krista Case argues that as agents gain access to sensitive data and critical workflows, companies need to know whose identity those agents are using, what privileges come with it and where those controls are actually enforced. That is a lot more demanding than just locking down the app layer and hoping for the best.

Resilience is the other half of Oracle’s pitch. The company is pointing to validated backups, tested standbys, disaster recovery and high availability, plus Zero Data Loss Recovery and its Globally Distributed AI Database with Raft-based replication and automated failover. The promise is blunt: keep applications running even if infrastructure components, availability zones or entire sites go down. That is the kind of pitch enterprises listen to when the fear is not just attack, but machine-speed mistakes too.