Oracle Targets Faster Patching as AI Threats Intensify | Cloud Wars Live

**Bob Evans** (0:08)
Hello, everyone, and welcome to this special edition of Cloud Wars brought to you by our friends at Oracle. Cyber attackers are using AI to discover and exploit vulnerabilities at a speed that traditional security processes were never designed to address.
To discuss that profound challenge, I'm joined by Oracle head of database security, Vipin Samar, to outline Oracle's expanded AI security strategy, which includes free tools to help Oracle AI database users protect against these AI-powered attacks. Vipin, welcome to Cloud Wars. Great to see you.

**Vipin Samar** (0:41)
Thank you, Bob. Great to be here.

**Bob Evans** (0:44)
So Vipin, there is just widespread urgency around the industry to help customers prevent these AI-driven attacks. Did that play a big part in the timing of your announcement?

**Vipin Samar** (0:56)
So there are two big developments that we have been seeing. First is certainly, as you have heard, people are rolling out agentic AI to improve efficiencies, effectiveness, and all of that.
But they are also realizing that AI can hallucinate, it can bring data it is not supposed to bring. So there are lots of concerns around it, right? And if you think about it, I mean, it makes sense because AI needs data, a lot of data.
Some of it could be sensitive actually, right? And how do you control access to it? Because people want to have proper controls on which they have full oversight. But AI makes the rules. What it should access, not access, it saw a lot of sensitive data there. Like Bob, the way I think about this is that it's like a very powerful insider.

**Bob Evans** (1:48)
Yeah.

**Vipin Samar** (1:49)
Broad access, broad reach, rules that it defines for itself. So it's not surprising that people are a bit concerned. So it's a big problem right now, right? So that is the first problem. The second one has been the advent of the new fast LLM models, which are becoming very powerful by the day actually, right? And we have all heard about, they write code and great code now, but they can debug code. They can find holes in that code, and if prompted, they can tell you how to exploit that hole. And that's very dangerous. Earlier on Bob, the hackers, they were the special hackers, and they'll have training and they'll take months and whatever time frame to find a hole, exploit it. But now with the new models, anybody and everybody can become a hacker. I mean, that is the real world, and they can find any hole anywhere on the stack, and it's a problem. They can not just find holes looking at source code, they can look at your binary.
And they can figure out where the holes are. And the worst part is this though, so all vendors, when they fix those holes, they give you patches.
Now, they can decompile those patches and understand the difference between the original and the patch, which means it's even easier to build pointed exploits to go after. So, that's the world in which we live in, where, where agent AI deployment has to happen. The new models are so powerful that anybody and everybody can become a hacker and hack into your IT systems. That's the problem.

**Bob Evans** (3:41)
So yeah, Vipin, unfortunately, I think we're hearing the same thing from a lot of customers. These threats are more powerful, more imminent, more automated and more dangerous. So in light of those new threats, what sort of the new Oracle approach that you're sharing with customers to help them ward off these new AI-powered attacks?

**Vipin Samar** (4:02)
So Bob, we have a two-pronged strategy here. One is a secure source will secure your data, will secure at speed because threats are coming at speed and secure through resilience.
In case something were to happen, we have to be able to come back up and running because we can't really take any downtime.
Now, that's the first one.
Second approach that we're taking is to remove the adoption barriers of deploying patches, for example, and really making our tools available, our industry-tested, hardened tools are available to everybody, to all the thousands and thousands of customers so that they can be ready when a patch is shipped, they can deploy the patch immediately and become secure and safe. Otherwise, the bad guys would be able to get in. Then other security tools as well, so that they have full visibility into the risk that they carry. Otherwise, right now, hackers know everything about you, but we have no idea where your risks are, hackers do. We want them to give the same tools so that they understand where the risks are and be able to protect themselves. That's our two-prong strategy here.