Google Cloud Next 2026: Agentic AI to Become ‘Operational Necessity’ for Security

“We can’t just take our old cyber playbooks and use them; we have to update those playbooks. A lot of the principles still apply, but we need to update them in a number of important ways. We need a new infrastructure that is AI-native,” said Francis deSouza, COO and president of security products at Google Cloud, during a session highlighting the security focus at the tech giant. 

WATCH: Get a view of the Expo Floor at Google Cloud Next 2026. 

While enterprise security teams are evolving with the help of AI, that also means malicious actors have those capabilities to mount their offensives, from fully agentic attacks to older tactics enhanced with newer tricks (such as multimodal phishing attempts). Businesses need to respond to threats at “machine speed,” deSouza said. 

At Google, the focus will be on continued investments in visibility into the threat landscape, deSouza said, refining its full-stack AI infrastructure and co-engineering security projects with AI infrastructure and models so that customers can take advantage of the latest models as soon as they’re released. The company is “shifting left” to build security into its services. 

Target Senior Vice President and CISO Jodie Kautt discussed why the retail giant had to pivot on its approach to security. 

“Typically, at Target, we were really a build shop, and that served my team really well for a long time,” Kautt said, noting that the company has a deeply technical team, dozens of security patents and a Cyber Fusion Center on threat analysis and detection. But, for the future, that approach needed adjustment. 

READ MORE: Expanding AI agent adoption requires a culture shift. 

“What we’ve pivoted to was a build and partnership strategy. We’re going to still build when we have unique problems to solve for Target or to solve for retail,” she said. “But we’re going to partner where we can extend our capabilities with organizations like Google.” 

Kautt noted that the partnership has helped Target automate key processes and reduce the time needed for triage. Analysts now have context in one pane of glass that they didn’t have before, and the security team has customization options. 

The teamwork of security is not only about sharing intelligence but also for improving processes and capabilities. “We have to be linking arms harder than we’ve ever had to before. Security is a team sport where we can take our own strong cyber capabilities and extend our teams to critical partners like Google and others,” Kautt said.

Rethinking SOCs With AI Agents 

During a separate session, Spencer Lichtenstein, group product manager at Google Cloud, highlighted how more advanced cyberattacks, which once took a matter of weeks to orchestrate, now take only a few days or hours, increasing the asymmetry between offensive-defensive responses. 

“Speed has increased, costs have gone down, and your attack surface has grown,” Lichtenstein said. 

That’s why SOCs need to shift their thinking, and Google Cloud has responded by improving their services to address previous workflow integration issues; include better explainability of what AI agents are doing in their security roles; add metrics measuring usefulness and quality for a reliable feedback loop; and make contextual understanding of an organization readily available for the agents. 

DISCOVER: Strategic partnerships turn AI ambition into action. 

“A lot of our focus for the future will be on adding elements of customization and contextual awareness to ensure that customers can tune the knobs the way that they need to in order to operationalize this effectively,” Lichtenstein said. 

Later, he was joined by Eric Herscovich, senior security engineer for detection and response at Shopify, to discuss how the multinational e-commerce company has transformed its SOC by going “all in” on agentic AI in the past year. 

Click the banner below to keep up with the latest BizTech event coverage. 

“Using agentic AI in the SOC is becoming an operational necessity,” Herscovich said, adding that the company is building out an agentic framework and that analysts are using agents to investigate, speed up workflows and reduce time to triage alerts. Fostering a culture of sharing among analysts has helped with adoption at Shopify, as analysts share their lessons through a Slack channel. 

GET THE DETAILS: How to deliver cyber resilience with expert partners. 

He noted that human analysts’ roles will transform to becoming orchestrators of AI agents, ushering in the next generation of detection response. A good first step for companies still getting their feet wet would be with a triage agent. 

“I think it’s really important that you start experimenting with it because everyone uses AI differently, every SOC will use it differently, each analyst within your SOC will use it differently. That human-computer interaction is very individualized,” Herscovich said.