Check Point brings open research, objective benchmarks and customer-controlled runtime protection to the industry initiative introduced by NVIDIA. 

AI is rapidly changing how organizations build, operate, and protect their digital environments. As a leader in the global cyber community, we believe in the power of collective intelligence. We are proud to join the Open Secure AI Alliance as an inaugural member, working alongside leaders from across cyber security, cloud computing, enterprise software, AI, and the open-source community. 

Introduced by NVIDIA, the alliance is creating a collaborative foundation for developing and sharing open technologies that advance AI safety and security, and will help organizations identify, remediate, and responsibly disclose vulnerabilities through open and inspectable technologies. 

Joining the alliance is a natural extension of work already underway. Our collaborations with NVIDIA and the broader research community have produced public frameworks, datasets, evaluation harnesses, reproducible research code and open benchmarks designed to make AI systems safer to build and deploy.  

Check Point brings open AI security research and objective security measurement to the Alliance as key differentiators.

Advancing Open AI Safety Research

The AI transformation is in full motion, with models, agents and AI-enabled systems are becoming part of the operating fabric of the enterprise. They can interact with sensitive data, invoke tools, access infrastructure and make decisions with increasing levels of autonomy. 

This shift introduces risks across the entire AI stack, including models, orchestrators, tools, applications, identities, data, guardrails, and runtime environments. Protecting that stack requires visibility and coordination across technologies and organizations. 

The Open Secure AI Alliance recognizes the important role that open models, harnesses and tools can play in this effort. Open technologies give defenders greater ability to study, test and adapt defensive capabilities while maintaining control over their data and infrastructure. They also support community-driven defense across multi-vendor environments, helping reduce reliance on any single system or provider. 

Organizations need security that works across their existing environments rather than requiring every capability to reside within one closed platform. Open, interoperable security allows intelligence to move between systems, protections to be coordinated and the broader community to benefit from shared learning. 

Advancing AI Security Through Open Research 

Researchers from NVIDIA and Check Point co-authored the paper introducing NVIDIA’s Agentic Safety and Security framework and collaborated on the public safety and security capabilities released through NVIDIA NeMo Agent Toolkit. Together, this work produced an open agentic safety dataset containing more than 10,000 standardized agent traces, including thousands of security attack and defense traces generated by Check Point for research and benchmarking. The NeMo Agent Toolkit implementation enables developers to incorporate end-to-end agent red teaming, runtime verification, and security testing into agentic AI workflows.  

We have also contributed objective benchmarks that help the broader community measure AI security. B³: Breaking Agent Backbones, developed with the UK AI Security Institute, evaluates how backbone LLM selection impacts agent security using nearly 200,000 human adversarial attacks.  We have also released additional open research assets, including our Apache-2.0 licensed adaptive LLM defense research and Intent Augmentation, public research code for generating and analyzing adaptive prompt-injection data. 

Turning Shared Knowledge Into Collective Protection 

Cyber security works best when knowledge is shared. A vulnerability discovered by one researcher can help protect thousands of organizations. Threat intelligence gathered in one environment can help defenders recognize attacks elsewhere. Shared frameworks, datasets and standards can raise the security baseline for an entire ecosystem. 

This model becomes even more important as AI systems grow more autonomous. Open collaboration can help transform individual discoveries and lessons from AI security incidents into protections that benefit the wider community. 

By joining the Open Secure AI Alliance, we look forward to contributing datasets, benchmarks, red-teaming methods, open research assets, and enterprise runtime-security experience to a shared mission: working toward unlocking the innovation gains that come from AI, safely and securely.  

Check Point’s participation reinforces a clear commitment: helping organizations adopt AI with confidence while advancing security practices the broader ecosystem can inspect, test and improve. Through the Alliance, Check Point will help move AI security from shared research into practical protection for enterprise environments. 

Read NVIDIA’s Open Secure AI Alliance announcement.