Netskope expands Google Cloud AI Guardrails partnership

Netskope has expanded its partnership with Google Cloud around its AI Guardrails security product, aimed at securing generative AI and autonomous AI agents running on Google Cloud.

The offering uses Google Cloud Tensor Processing Units and Vertex AI to apply safety and moderation checks to AI workflows while they run. It is designed for organisations deploying large-scale AI systems and agent-based tools that act independently.

The announcement comes as companies increase spending on artificial intelligence and shift from chatbot-style systems to software agents that can carry out tasks, call external tools and interact with model context protocol servers. That shift has widened the security challenge, as these systems can exchange data with multiple services and take actions with less direct human oversight.

How it works

AI Guardrails sits inside a customer's Google Cloud environment, where prompts and responses can be scanned locally rather than sent outside the tenant. Netskope said this is intended to help customers keep control of sensitive information and meet regulatory and data residency requirements.

According to Netskope, the system performs real-time checks on content categories including hate speech and discrimination, criminal activity, suicide and self-harm, piracy and copyright, and requests for sensitive data. It also examines interactions between AI agents and the tools they use to identify indirect prompt injection and other attempts to manipulate system behaviour.

The product is built to detect prompt injection, jailbreaking and other threats specific to large language models. Netskope added that it can also identify unintended recursive loops and malicious ghost commands that could drain computing resources or trigger unauthorised changes inside cloud environments.

Enterprise demand

Security suppliers are increasingly positioning themselves around a new stage of enterprise AI adoption, as companies move beyond experimentation to operational use. In that environment, customers are under pressure to show that AI systems can be governed in line with internal policy, sector regulation and broader model risk frameworks.

Netskope said detections from the product are mapped to frameworks including MITRE ATLAS and the OWASP Top 10 for LLMs. This is intended to give security teams a record of how AI-related risks are identified and managed.

The company also linked the product to compliance demands around GDPR, HIPAA and the EU AI Act, arguing that local inspection and policy enforcement can support customers that need tighter control over regulated data. These issues have become more prominent as businesses test AI systems on customer information, internal records and proprietary documents.

Google Cloud's role in the partnership centres on hardware and AI services infrastructure. Netskope said TPUs allow security checks to run in line with model speed, while Vertex AI supports real-time moderation.

For Google Cloud, the partnership adds another security layer around AI services customers may already be using for model deployment and inference. For Netskope, it provides a route into enterprise projects where buyers want security tools embedded close to the models and data they are trying to govern.

Vineet Bhan, Director of Security and Identity Partnerships at Google, commented on the arrangement.

"Netskope and Google Cloud's collaboration helps bring the best of AI innovation from Google Cloud and enterprise security from Netskope," Bhan said. "Netskope One AI Guardrails works with Vertex AI and TPUs to enable the secure deployment of both generative AI and autonomous agents. Together, we are helping organisations build the future of their business on a foundation of trust."

Netskope framed the product as part of its wider AI security portfolio, intended to cover models, applications, agents, tools, users, devices and data across cloud environments. It said the latest work with Google Cloud is meant to address security concerns that arise when AI systems act more autonomously and interact with other software in real time.

"As a global leader in AI-ready security and networking, Netskope provides critical capabilities required for the safe, high-performance adoption of modern AI workflows," said Sanjay Beri, Chief Executive Officer and Co-Founder of Netskope. "We are proud to collaborate with Google Cloud to help ensure that customers can confidently pursue their AI goals."