The newly developed platform, in partnership with NVIDIA, is dedicated to securing AI models, workloads, and agentic applications utilized in enterprise settings, all while maintaining optimal performance. The company has announced that AI Cloud Protect is now accessible for on-premises enterprise deployment and has been validated on NVIDIA RTX PRO Servers.
Built upon NVIDIA’s BlueField technology, this platform empowers organizations to safeguard AI model development and inference workloads in both data centers and cloud environments.
“As enterprises strive to foster AI-driven innovation, they cannot afford to have blind spots,” stated Nataly Kremer, Chief Product Officer at Check Point. “In collaboration with NVIDIA, we are ensuring that AI factories are secure by design—safeguarding models, data, and infrastructure without hindering innovation.”
The emergence of AI has introduced new and intricate risks for enterprises. According to data from Check Point, one in every 80 generative AI prompts discloses sensitive information, while Gartner indicates that nearly one-third of organizations experienced an AI-related security incident in the previous year, which included issues ranging from prompt manipulation to infrastructure attacks.
AI Cloud Protect was specifically designed to confront these vulnerabilities directly. Operating on NVIDIA BlueField-3 DPUs, it provides comprehensive protection without utilizing CPU or GPU resources, effectively eliminating the traditional compromise between security and performance.
Its network-level defense system thwarts unauthorized access, data poisoning, and model exfiltration. At the host level, it utilizes NVIDIA’s DOCA Argus framework for direct memory access, granting complete visibility into active processes on AI nodes to identify and prevent malicious workloads, even within large language models that have been downloaded.
Also Read: Court Orders Takeover of Nestoil Headquarters Over Alleged $1 Billion Debt
“Security is crucial for the forthcoming generation of AI infrastructure,” remarked David Reber, Chief Security Officer at NVIDIA. “NVIDIA is collaborating with Check Point to incorporate BlueField acceleration and the NVIDIA DOCA Argus runtime security framework into the AI Cloud Protect platform.
The forthcoming NVIDIA BlueField-4 DPU will also benefit from this solution, which is expected to deliver six times the computational power and double the network throughput, thereby facilitating quicker and more scalable AI operations.
In addition to AI Cloud Protect, Check Point’s AI security suite features the CloudGuard Web Application Firewall (WAF) integrated with Lakera, specifically designed to mitigate prompt injection and jailbreak threats within enterprise AI applications.
Another offering, Infinity GenAI Protect, assists organizations in overseeing generative AI usage, implementing policies, and safeguarding sensitive data from exposure across teams. Collectively, these solutions are intended to provide comprehensive protection throughout the AI ecosystem, from infrastructure to the user level.
Currently, the company is testing AI Cloud Protect with selected financial services firms and partners such as World Wide Technology (WWT), with an emphasis on securing data centers that support the development of large language models.
“As enterprises scale their AI server factories, the integration of Check Point’s AI Cloud Protect with NVIDIA BlueField acceleration offers enterprise-level protection for sensitive AI workloads, from model training to inference, without sacrificing the performance required by modern AI applications,” stated Chris Konrad, vice president, Global Cyber, WWT.
Check Point is tackling issues related to data leakage, model manipulation, and infrastructure threats, while empowering organizations to innovate with confidence during this era of intelligent computing.
