Oracle Deploys AI-Powered Security Strategy as Frontier Models Expose Database Vulnerabilities

The emergence of frontier artificial intelligence models capable of identifying software defects at scale is pushing database makers to implement larger and more frequent security updates. Real application testing has become critical infrastructure for ensuring that these rapid patches do not introduce regressions in production environments. Simultaneously, the same AI systems are circumventing traditional security controls embedded in application code.

Oracle Corp. is tackling this challenge through two distinct initiatives: strengthening its codebase against external threats and relocating access controls to the database layer itself. The company's effort began when it obtained early access to an advanced AI model and tested it against its own systems, according to Juan Loaiza, executive vice president of Oracle Database Technologies.

Back in March, we got early access to the Mythos model and we ran it against our code and it started finding all these security vulnerabilities. It's turning up some very serious security issues. And so we said, 'Hey, this is a code red moment. The house is on fire.'

Juan Loaiza, Oracle

Real application testing meets bigger, faster patches

Oracle's primary response to AI-discovered vulnerabilities is being delivered through its quarterly patch cycles, which have expanded significantly in scope. A security update released earlier this year resolved more than 1,400 vulnerabilities across Oracle's product suite. Customers face pressure to automate patching and maintain current versions, yet each substantial patch introduces the risk of breaking existing systems.

We have something called Real Application Testing where we can capture real workload and run it to make sure that your workload is not affected by these patches. It's not quite as simple as just apply because the applications that people run are extraordinarily complex. They run entire countries, entire industries.

Juan Loaiza, Oracle

While patching addresses external attackers, the internal deployment of AI systems presents a distinct security challenge. Historically, applications have enforced access controls at the user level, but AI systems now generate much of this code or bypass it entirely to access data directly. Oracle's answer is Deep Data Security, which began general availability in April and transmits an authenticated user's identity as an encrypted token down to the database layer. The company has since expanded these authorization mechanisms to encompass remote databases, lakehouse architectures and associated data tables.

The new world as I see it is all these end-user security rules have to be enforced where AI can't bypass them, which means all the way down in the data. The AI agent has to get your identity. It's acting on behalf of you. It can't leak the data if it can't see the data.

Juan Loaiza, Oracle

Business logic constraints require the same treatment. Traditionally, application code has prevented credit card transactions exceeding a customer's available balance, but an AI agent tasked with enforcing this rule may disregard it. Oracle is introducing declarative mechanisms for specifying such constraints, along with SQL Assertions that validate whether modifications executed through SQL comply with defined rules.

https://www.youtube.com/embed/05nqgG1zbfE?list=PLLTiNyEQRCms

There's credit limits, there's all sorts of things that are business rules that can't be violated. And again, if we move those all the way down into the data, then no matter how you access the data, what mechanism you use, how AI is used, it cannot violate those rules.

Juan Loaiza, Oracle