At Oracle AI World in Las Vegas, Nevada, OpenAI CEO Sam Altman has rallied the security community and called for cyber pros to keep focus, saying the only way through the current tsunami of patches is through.
In conversation with Oracle co-CEO Clay Magouryk during the day two keynote, Altman aligned with Magouryk that AI is not necessarily creating the problem but rather helping discover the myriad of security issues and flaws that have slowly developed over the past 20 to 30 years, and longer in some cases.
“I think the only way through this is to go and find all these issues and fix them,” said Altman. “In fact, I think we’ll need new ways to fix them. It’s maybe too hard to patch every bug one by one across the entire supply chain that we’re all dependent on. I think we’ll have to look at new kinds of services that are constantly monitoring traffic and not letting these things be exploited.
“But I find myself shocked at just how many vulnerabilities these AI systems are finding and what it’s going to take – just the incredible, Herculean effort that it’s going to take to get ahead of this.”
Asked what advice he would give to cyber pros at this point, Altman said: “If I could recommend only one thing to a company right now to do with AI in the very short term … it would be to really understand the security threat landscape from AI and mitigate against that.”
He added: “I do realise that this is probably somewhat annoying for me to say and you’re like, well, you’re creating the problem.”
Magouryk said: “I’m glad that we are now able to discover [flaws] with AI rather than waiting for some unknown hacker to let us know about them.”
Cyber mesh
During the keynote session, Magouryk added colour to new security initiatives trailed earlier at the event, with the formal announcement of the new OpenAI-powered Oracle AI Security Mesh, designed to complement the existing Oracle Cloud Infrastructure (OCI) Identity Assurance – a biometric human identity verification tool – and OCI Zero Trust Packet Routing – a network access policy enforcer.
Oracle proposes that this mesh service will enable its enterprise customers to strengthen protections across both applications and infrastructure whether they are running on-premise, in OCI, or in someone else’s cloud.
The combination of OpenAI’s latest frontier models with Oracle’s testing tools and contextual knowledge of customer systems will supposedly let users find and fix flaws by reasoning through application behaviour, investigating possible weaknesses, or assessing fixes.
Any changes or fixes applied will be human-approved with agents designed to operate with distinct identities, task-scope, and just-in-time permissions, and independent monitors will be on hand to detect and stop any agentic activity that ranges across approved boundaries.
Going deep with OpenAI
But Oracle’s tie-up with OpenAI does not stop with cyber security. The software supplier also today announced a new collaboration with the AI giant to marry up its ChatGPT and Codex models to Oracle’s applications, data and cloud infrastructure to put AI to work across customer systems and operations.
An expanded product suite will bring this approach to life, said Oracle, by helping people work across systems, build and deploy applications, get on top of cyber issues. It described an “integrated foundation to build and run companies”.
“OpenAI is transforming how enterprises solve business problems, from chat, to coding, to analytics, to autonomous agents,” said Magouyrk.
“With this new product suite, we are uniting their transformative AI capabilities with Oracle applications, data, and cloud infrastructure – technology that businesses around the world rely on to run their most critical operations.”
