Thales is expanding its partnership with Google Cloud. The Thales AI Security Fabric is being integrated with the Gemini Enterprise Agent Platform. This gives organizations real-time security, governance, and visibility into interactions among users, agents, models, and tools.
The partnership addresses a growing challenge as companies deploy AI agents in production. Unlike traditional applications, agents do not follow a predefined workflow. They reason, plan, and make decisions independently. They also communicate with other agents. According to Thales, this creates new attack surfaces that traditional security architectures struggle to keep up with.
“Companies are transitioning from AI assistants to AI agents that can take independent action, make decisions, and communicate with business-critical systems. This requires a fundamentally different approach to security,” says Eva Rudin, Senior Vice President of Cybersecurity Products at Thales.
Thales cites prompt injection, the leakage of sensitive data, unsafe outputs, unauthorized actions, and increasingly complex agent-to-agent interactions as risks.
Frameworks around agents
The AI Security Fabric establishes frameworks and controls around agents. These are designed to keep agents within their authorized boundaries, restrict data access, and block inappropriate actions in real time. According to Thales, this security layer applies across all interactions. Organizations can use it to enforce policies, detect AI-specific threats, and continuously monitor agent behavior. The platform also offers centralized governance and compliance features.
Vineet Bhan, Director of Security and Identity Partnerships at Google Cloud, says that customers need security and governance built in by default, now that agents are linked to critical data and systems. He believes the partnership with Thales offers additional capabilities to protect agentic AI workflows.