Google Cloud Fortifies AI and Cloud Defenses with New Security Innovations at Summit 2025
At the Google Cloud Security Summit 2025, the company unveiled a comprehensive suite of new security capabilities designed to bolster defenses for artificial intelligence (AI) agents and cloud workloads. This initiative underscores Google Cloud's commitment to making security an enabler of business ambitions, focusing on automating compliance, simplifying access management, and expanding data protection for AI initiatives. The announcements reflect a dual strategy: protecting organizations' AI capabilities while simultaneously employing AI to strengthen their overall security posture.
Securing AI Innovation with AI Protection
A cornerstone of the summit's announcements was the introduction of "AI Protection," a new suite of security features engineered to mitigate risks across AI workloads and data, irrespective of the underlying platform. This offering is integrated into Google Cloud's Security Command Center (SCC), providing a centralized system for AI security management alongside other cloud-related risks. AI Protection aims to offer businesses a consolidated view of their AI standing, enabling them to manage risks and detect threats proactively.
The core capabilities of AI Protection are threefold: AI Inventory Discovery, which identifies and assesses AI assets for vulnerabilities; AI Asset Security, which implements controls, policies, and guardrails to secure AI resources; and Threat Management, which provides mechanisms for detection, investigation, and response to AI-related threats. This holistic approach is designed to help teams comprehensively manage AI risk throughout the AI lifecycle.
Google Cloud's Sensitive Data Protection (SDP) Enhancements are being extended to Vertex AI datasets, facilitating the automatic discovery and classification of sensitive training and tuning data. Once sensitive data is identified, AI Protection leverages SCC's virtual red teaming capabilities to pinpoint potential attack paths on AI systems and suggest remediation steps. Furthermore, "Model Armor," a core component of AI Protection, is now generally available. Model Armor is designed to defend AI agents against threats such as prompt injection, data leakage, jailbreaking, malicious URLs, and offensive content. It can be integrated into applications via REST API, Apigee, and soon Vertex AI, offering real-time safeguards for agent prompts and responses.
AI Protection also operationalizes security intelligence and research from both Google and Mandiant. SCC's existing detectors are enhanced to uncover initial access attempts, privilege escalation, and persistence efforts related to AI workloads. New detectors, informed by the latest frontline intelligence, are slated for release to help identify and manage runtime threats, including foundational model hijacking.
The Agentic SOC Vision and Enhanced Operations
Google Cloud also elaborated on its vision for an "agentic SOC" (Security Operations Center). This concept envisions a system where AI agents coordinate their actions to achieve shared security goals, automating tasks such as alert triage, investigation, and response. An Alert Investigation agent, currently in preview within Google Security Operations, exemplifies this vision. This agent can autonomously enrich events, analyze command-line activity, and build process trees, emulating the practices of Mandiant's human analysts. This automation aims to reduce manual workloads for security teams and accelerate response times.
Complementing these operational enhancements, Mandiant Consulting is expanding its role in AI security. New services are being offered for AI governance frameworks, AI environment hardening, and AI threat modeling. These services are designed to help organizations assess risks through adversarial testing and implement robust defense mechanisms before deploying generative or agentic AI.
Updates Across the Google Cloud Security Ecosystem
Beyond AI-specific protections, Google Cloud announced several updates to its broader security portfolio. The Security Command Center is introducing new previews, including Compliance Manager to unify compliance workflows, Data Security Posture Management integrated with BigQuery Security Center for enhanced data governance, and new risk reports powered by virtual red team technology to identify cloud security issues.
In Identity and Access Management (IAM), upcoming features include "Agentic IAM" for provisioning agent identities across cloud environments, a Gemini-powered IAM role picker to recommend least-privilege roles based on described actions, and re-authentication for sensitive actions to prevent unauthorized access. These updates aim to simplify access management and strengthen identity controls for both human and non-human identities.
Data security enhancements include expanded coverage in Sensitive Data Protection to monitor Vertex AI, BigQuery, and CloudSQL assets, inspect images for sensitive data, and detect various context models. Cloud Key Management System Autokey is now generally available, streamlining the onboarding process for customers requiring customer-owned encryption keys.
Network security updates include expanded org-scope tags in Cloud NGFW with hierarchical support, preview support for RDMA networks to enable Zero Trust networking for high-performance computing, and expanded Cloud Armor Enterprise features such as hierarchical policies and ASN-based threat intelligence for enhanced centralized control and protection.
Balancing Innovation with Responsibility
The innovations announced at the Google Cloud Security Summit 2025 underscore a commitment to enabling organizations to innovate with AI while ensuring robust security, compliance, and privacy. By automating compliance, simplifying access management, and expanding data protection for AI workloads, Google Cloud aims to empower security teams and foster resilient defenses. The company emphasizes that by using AI to empower defenders and meticulously securing AI projects from inception to deployment, Google Cloud provides a comprehensive foundation for organizations to thrive in the evolving AI landscape. The overarching message is that security should be an enabler, not a barrier, to business ambitions in the age of AI.
AI Summary
Google Cloud announced a significant expansion of its security offerings at the Security Summit 2025, focusing on protecting AI workloads and agents while leveraging AI to enhance defenses. Key among the announcements is "AI Protection," a new suite of features integrated into Security Command Center (SCC). AI Protection aims to mitigate risks across AI workloads and data, regardless of the platform used, by providing capabilities for discovering AI inventory, securing AI assets, and managing threats. This includes automated discovery of AI agents and Model Context Protocol (MCP) servers, enhanced in-line protection against threats like prompt injection and data leakage through "Model Armor," and new threat detections informed by Mandiant and Google intelligence. The company also detailed its vision for an "agentic SOC," where AI agents automate tasks like alert triage and investigation, exemplified by the Alert Investigation agent in Google Security Operations. Mandiant Consulting is expanding its AI security services, offering expertise in governance, environment hardening, and threat modeling. Updates to Google Unified Security include new dashboards in Security Operations and experimental AI-driven features in SecOps Labs. Chrome Enterprise is extending protections to iOS and Android to combat shadow AI risks. On the platform side, Security Command Center is gaining Compliance Manager, Data Security Posture Management, and Risk Reports. Identity and Access Management will introduce Agentic IAM, a Gemini-powered role picker, and re-authentication for sensitive actions. Data security enhancements include expanded coverage in Sensitive Data Protection for Vertex AI and other AI-related assets, along with updates to Cloud Key Management System Autokey. Network security sees improvements in Cloud NGFW and Cloud Armor Enterprise. The overarching theme is enabling organizations to confidently build and deploy AI securely, compliantly, and privately, with security acting as a business enabler.