Google Cloud focuses on advancing and securing AI at Next 25

Google Cloud has released several security advancements at its ‘Next 25 conference, emphasising the integration of artificial intelligence (AI) and agents to enhance threat detection and response.

With this AI strategy comes the release of security products like Google Unified Security (GUS), a unified platform combining threat intelligence, cloud security and security operations.

Other announcements include the embedding of Gemini, Google’s AI model, across security products; an alert triage agent for analysing security alerts; a malware analysis agent for de-obfuscating and assessing code and Mandiant Threat Defence for integrated threat detection and response.

Additionally, there was also Chrome Enterprise enhancements like phishing protections and data masking for Android; Compliance Manager for policy enforcement, auditing and monitoring; Model Armor, an AI firewall integrated with Vertex AI for prompt and response filtering and the expanded Risk Protection Program, which now includes Beazley and Chubb for cyber insurance based on security posture.

Google product management executive Payal Chakravarty said the updated Compliance Manager for Google Cloud streamlines regulatory compliance.

In a demonstration, the alert triage agent was able to analyse security alerts, detect malicious traffic to a virtual machine (VM) and automatically associate it with an emerging threat actor that the Google threat intel team had been tracking in real time.

“GUS was able to recognise this risk in real time and took action,” she said. “Additionally, if you look at the recommendations that GUS made, the agent has [proposed] … hardening the AI model with Model Armor.”

Chakravatary said this shows that all prompts that are entered into the model are analysed in line in real time and malicious inputs are blocked before they reach it.

“Many of our enterprise customers are busy rebuilding all their applications to be AI-first, much like how people wrote their applications to be mobile-first during the mobile revolution many years back,” she said. “This comes with a new set of challenges.”

Google VP of cloud security engineering Brian Roddy said AI Protection for Google Cloud provides capabilities for managing risk for the whole the AI lifecycle within the context of cloud risk.

“Beyond this, a very cool innovation that’s gaining a ton of traction is what we call Model Armor, which is part of our AI protection … like an AI firewall,” he said.

“It takes Google’s safety technology and puts it into a proxy that can inspect both the prompts and responses and [allows] you to apply true safety and security controls.

This works across a broad range of models. It’s not just Gemini; it’s all the major leading models.”

The security product releases come as Google Cloud also revealed a number of changes to its partner go-to-market approach and launched a new marketplace for AI agents at ‘Next 25, among other updates.