Check Point Software Technologies Ltd. (NASDAQ: CHKP) announced it is partnering with NVIDIA to deliver AI Cloud Protect, a security solution for enterprise AI infrastructure. The product is now available for on-premises deployments and has been validated on NVIDIA RTX PRO Servers.

AI Cloud Protect runs on NVIDIA BlueField-3 DPUs and is designed to secure AI model development, applications, and inference workloads without impacting performance. The solution provides network-level protection and host-level security using NVIDIA DOCA Argus technology to monitor processes on AI nodes.

According to Check Point data, one in every 80 generative AI prompts exposes sensitive data. A Gartner report cited in the announcement found that 32% of organizations experienced AI attacks involving prompt manipulation in the past year, while 29% faced attacks on their generative AI infrastructure.

The solution addresses security challenges in AI deployments including data poisoning, model exfiltration, and unauthorized access. It operates across network, host, and application layers while leveraging NVIDIA BlueField acceleration to avoid consuming CPU or GPU resources.

Check Point is currently piloting AI Cloud Protect with financial services customers and strategic partners including World Wide Technology. The company also plans to support the upcoming NVIDIA BlueField-4 DPU, which is expected to offer increased compute power and network throughput.

The announcement includes quotes from Check Point Chief Product Officer Nataly Kremer and NVIDIA Chief Security Officer David Reber emphasizing the importance of security for AI infrastructure deployment.