At Security Summit 2025, Google Cloud introduced new cybersecurity tools that combine an AI agent approach with a unified security platform. The ambition is to dramatically speed up responses to attacks while strengthening the protection of the enterprise environment.
Agent-based security centres
So-called agent SOCs (Security Operations Centres), i.e. operations centres where autonomous AI agents operate alongside human analysts, have taken centre stage. For example, Google showed an Alert Investigation agent that can automatically enrich security events with context, analyze command lines and build process trees according to Mandiant’s methodologies.
The goal is to allow security teams to engage in more complex analysis and decision-making while agents take over the routine triage of incidents. This can reduce response time from hours to minutes, according to Google.
AI application security
The company has also expanded the Security Command Center with new features aimed at protecting AI agents and applications themselves. These include automatic discovery of AI services in the infrastructure, enhanced protection against prompt injection or sensitive data leaks, and new detection capabilities based on Mandiant technologies.
“Our vision is to offer developers a path that is not only secure, but also easy to use,” said Naveed Makhani, Security AI product lead at Google Cloud.
Extending the unified platform
Other new developments are aimed at strengthening Google Cloud’s unified security platform. These include the SecOps Labs testing environment for new AI-driven features, the Agentic IAM service for automated assignment of identities and permissions to AI agents using the Gemini model, improved data protection in Sensitive Data Protection and Cloud Key Management Autokey, and new features for Chrome Enterprise that enhance mobile security.
Competition and challenges
While Google is building on its position in cloud services and cybersecurity, competitors are taking similar steps. Both Microsoft and AWS are investing in connecting security platforms with AI assistants. The challenge for all players is to strike a balance between speed and security. AI can help detect an attack faster, but it also opens itself up to new attack vectors, such as misuse of models or false decisions in incident assessment.
Impact on the industry
For companies in the engineering, energy or logistics industries, this shift could be significant. Manufacturing businesses are facing increasing pressure to digitise and connect systems, increasing the risk of cyber threats. Automated SOCs with AI agents can reduce reliance on scarce human experts and increase the resilience of critical infrastructure.
The first pilot projects are emerging in Europe. Energy companies are testing real-time monitoring of distribution networks and manufacturing plants are using AI to detect non-standard behaviour in control systems. It is in industries with high levels of automation and critical operations that these solutions are expected to have the greatest impact.
From an industry perspective, it’s not just about the cybersecurity of IT systems, but also about protecting production lines, power grids and transportation hubs. Introducing AI into security operations is therefore becoming an important step towards ensuring that digitalization is not also the Achilles heel of industrial companies.
Source
