Oracle announces AI Security Mesh, a security tool powered by OpenAI models to find and fix vulnerabilities
IA en un minuto newsroom · Editor: Jon Elgezabal
In 30 seconds
Oracle is launching a managed service, AI Security Mesh, that sets OpenAI cybersecurity models to hunt for and fix flaws. It covers OCI, other clouds and customer-run sites, and adds those models to the testing tools of Oracle. Oracle lets the customer set where testing happens and how far it goes, and no availability date is given.
Oracle is leaning on OpenAI to strengthen security in the cloud. The company has unveiled Oracle AI Security Mesh, a managed offering built on OpenAI's cybersecurity models, aimed at spotting and repairing weaknesses in applications and infrastructure ahead of attackers. It sits inside a wider set of announcements made with OpenAI, set out in Oracle's release on PR Newswire, and Unite.ai dates the launch to October 27, at Oracle AI World in Las Vegas.
Oracle says the service reaches OCI, rival clouds and customers' own data centers. The OpenAI models are paired with Oracle's own test tooling, outside validation and test infrastructure, sparing customers from assembling and running defensive and offensive AI of their own. Unite.ai adds that it complements OCI Identity Assurance, a biometric identity check, and OCI Zero Trust Packet Routing, which applies network access policies; it also reports that nothing is fixed without the customer's sign-off.
Customers choose where tests are permitted and how far they reach. Oracle looks after the protections meant to keep tests from knocking systems offline or disrupting operations. The release states no availability date or price.
Why it matters · analysis and opinion
Actively hunting for security flaws is now offered as a managed service backed by AI models, which brings that kind of review within reach of companies without a security team of their own. The design stresses control: the customer sets the scope of testing and, according to Unite.ai, no fix is applied without its approval, which matters when AI agents probe real systems. Questions remain open: Oracle has not published a price or a date, and the claims about speed come from the company itself, without an independent evaluation. For a business that already hosts data on Oracle or on other clouds, it is worth asking the provider how the scope is limited, what gets logged and who answers if a test causes an outage.
Source: PR Newswire · Written with the help of AI: how we make the news
