Thales has expanded its partnership with Google Cloud to help enterprises secure autonomous AI agents, as organisations move from AI assistants that answer questions to agents that can act on their own within business systems.

Thales‘s AI Security Fabric is now integrated with Google Cloud‘s Gemini Enterprise, applying security and governance controls across interactions between users, AI agents, models, enterprise data and tools in real time.

The collaboration is designed to address risks that come with agentic AI: prompt injection, sensitive data leakage, unsafe outputs, unauthorised actions and increasingly complex agent-to-agent interactions. Unlike conventional software that runs within predefined workflows, AI agents can reason, plan and act with other agents dynamically, which creates a broader and more unpredictable attack surface than traditional security tools were built to manage.

“Enterprises are moving from AI assistants to AI agents that can autonomously take action, make decisions, and interact with critical business systems. This requires a fundamentally different approach to security,” said Eva Rudin, Senior Vice President, Cybersecurity Products, at Thales. “By expanding our collaboration with Google Cloud, we are helping organizations build the necessary trust with security and governance as the foundation required for the next generation of enterprise AI.”

Keeping agents within their mandate

Thales gives an example: an AI agent tasked with settling insurance claims could exceed its mandate by pulling personal information from outside approved sources to influence a payout, creating privacy and compliance risks. Thales AI Security Fabric applies controls around agents to keep them within authorised boundaries, limit data access and block inappropriate actions in real time.

A unified layer across the AI stack

The security layer is meant to work across an organisation’s AI systems, helping enforce policy, detect AI-specific threats, maintain visibility into agent behaviour and support governance requirements as agents connect to more critical data and systems.

“Google Cloud helps enterprises move from experimenting with AI agents to deploying them securely across their businesses,” said Vineet Bhan, Director of Security and Identity Partnerships, Google Cloud. “As agents become more capable and connected to critical data and systems, customers need security and governance built into how they operate.”

Thales generated €22.1 billion in sales in 2025 and employs more than 85,000 people across 65 countries, with €4.5 billion a year going into research and development in areas including artificial intelligence, cybersecurity, quantum and cloud technologies.