Latching onto the key threat facing companies today, Oracle Corp. founder and Chief Technology Officer Larry Ellison announced a new cloud cybersecurity service that he promises will mostly automate the process of detecting and quashing data thefts.

Ellison (pictured), speaking at the database and business software giant’s annual OpenWorld conference in San Francisco, said the newly named Management and Security Cloud uses machine learning to detect anomalies and unusual patterns of use from activity logs in a wide range of databases, applications and cloud services, not just Oracle’s. “It takes the configuration information of all of your assets and unifies it, puts it in one place, keeps track of all of your users,” he said.

The announcement comes two days after Ellison introduced the next version of Oracle’s mainstay database, which he said will completely automate tasks such as making security patches and recovering data deleted by ransomware. The cybersecurity suite is not fully automated, but the two products are intended to work together to provide better security.

Threats are getting too numerous and too sophisticated to battle the current way, which usually involves means poring over log entries manually and then trying to do patches when there’s time, he said. In the case of Equifax, which had the critical records of 143 million Americans stolen, one patch took months, even after the hack was discovered.

“The current strategy is not working,” he said. “We are losing the cyberwar. Security in our data center, what is it, Job 10? We have to elevate the priority of security in our data center, because no one wants to be on the front page saying they lost their data.”

The new offering actually includes two integrated suites: the Oracle Management Cloud for applications and infrastructure monitoring and log analytics, and the Oracle Identity Security Operations Center, which provides a set of services such as access management and security monitoring.

All the log data from the clouds and on-premises computers, including databases and many applications such as HR, that customers use is ingested into this system. As it goes in, it’s “semantically enriched” with machine learning to note what kinds of logs or other data it is so it anomalies can be recognized automatically. Companies can use plain English queries to say, “Find me all the failed logins to the authentication system” or “Show me high-severity threats for my billing app,” he said.

Oracle is hardly the first to employ machine learning in security. They include cloud rivals such as Microsoft Azure, innumerable security startups and Splunk Inc., which allows customers to apply their own machine learning models to its log analytics. But Ellison touted the unified system that allows companies to “go directly from detecting a problem to fixing a problem.”

Photo: Oracle livestream