Lava Security’s AI Infrastructure Security Framework Becomes an Official OWASP Project

New York, USA, November 11th, 2026, FinanceWire

When organizations rent computing capacity from a neocloud provider, they are not simply relying on the security of an operating system or application. Their workloads depend on an extensive infrastructure stack that includes physical servers, firmware, networking equipment, management controllers, storage systems, and operational services.

That complexity introduces security questions that can be easy to overlook, particularly as AI infrastructure becomes available through shared environments.

One example involves bare-metal servers, where customers may receive root access to rented hardware. While that access allows organizations to control their computing environments, it can also create opportunities for changes to persist below the operating system. A compromised hardware boot configuration, for instance, could survive an operating system reinstall and affect the next customer assigned to the same server.

These are the types of scenarios examined in a new framework from Lava Security, which has officially become an OWASP project.

The company's Top 10 Neocloud and AI Data Center Security Risks framework was developed by Yakir Kadkoda, CTO and Co-Founder of Lava Security, and Michael Katchinskiy, Head of Security Research. The initiative aims to provide infrastructure providers, operators, and organizations running AI workloads with a shared reference for understanding and addressing security risks across the AI infrastructure stack.

Looking Beyond Traditional Cloud Security

The infrastructure supporting AI workloads differs from many conventional enterprise environments. Modern AI data centers bring together thousands of GPUs, high-speed networking, specialized management controllers, shared storage, and orchestration systems.

Much of this technology was originally designed for trusted high-performance computing environments. As commercial neocloud providers increasingly use these systems to serve unrelated organizations on shared physical infrastructure, the security requirements become more complicated.

The framework examines several areas where weaknesses could create broader exposure.

These include compromised server firmware, vulnerabilities in InfiniBand and RDMA networking technologies, inadequate tenant isolation, and insecure out-of-band management interfaces.

It also considers supply chain threats, data center facility systems, shared storage and AI artifacts, incomplete security certifications, centralized operational services, and delays in vulnerability patching.

The objective is to give organizations a way to evaluate risks that might otherwise fall between traditional infrastructure and application security practices.

"AI infrastructure introduces attack surfaces that traditional security approaches may not adequately address. Organizations need to understand what happens below the operating system, particularly when hardware, firmware, and management interfaces are shared across customers," said Michael Katchinskiy, Head of Security Research at Lava Security. "This framework helps make those risks easier to identify and provides practical guidance for addressing them."

Turning Technical Risks into Practical Guidance

A key part of the framework is its focus on potential attack scenarios and their operational implications.

Instead of simply listing vulnerable technologies, each risk category includes possible attack scenarios, real-world implications, and practical mitigation recommendations.

For an infrastructure operator, this can provide a starting point for reviewing how servers are configured, how management interfaces are protected, and how workloads are isolated from one another.

For organizations purchasing GPU cloud capacity, the framework offers a reference for understanding the security controls that should be considered when evaluating the infrastructure hosting their workloads.

This broader perspective is important because infrastructure components do not operate independently. Weaknesses in one area can affect other parts of the environment, particularly when hardware, networking, management systems, and workloads are interconnected.

"Securing AI infrastructure requires looking at the entire environment, not just individual components. Hardware, networking, management systems, and workloads are interconnected, and risks in one area can affect the security of the broader infrastructure," said Yakir Kadkoda, CTO and Co-Founder of Lava Security. "Developing this framework within OWASP gives the community an opportunity to contribute its expertise and continuously improve how these risks are understood and addressed."

Research Backed by Industry Practitioners

The framework was initially published in July 2026. Its development drew on disclosed vulnerabilities, incident reports, technical documentation, and Lava Security's research into AI infrastructure security.

It also incorporated feedback from more than 20 practitioners across neocloud providers, hardware OEMs, security companies, universities, and research institutions.

That range of contributors reflects the number of different stakeholders involved in securing AI infrastructure. Hardware manufacturers understand the technologies powering servers and networking equipment, while cloud operators and security researchers bring different perspectives on how those systems are deployed and exposed.

Bringing these perspectives together through OWASP gives the initiative a mechanism for continued industry participation.

An Open Development Model

The project will continue evolving through an open development process on GitHub. Security researchers, infrastructure engineers, hardware vendors, and operators can review the framework, challenge its assumptions, propose changes, and contribute additional research or mitigation recommendations.

Contributors will be credited in future releases.

The framework is also designed to evolve as the underlying infrastructure changes. New GPU architectures, networking technologies, management systems, and deployment models will introduce additional security considerations that may need to be incorporated into future versions.

For Lava Security, the OWASP transition establishes a foundation for making AI infrastructure security a broader industry discussion rather than a collection of disconnected research efforts.

The official OWASP project page and GitHub repository provide channels for those interested in reviewing the framework or contributing to its development.

As AI workloads become increasingly dependent on specialized infrastructure, understanding what happens beneath the operating system will remain an important part of the security conversation.

Contact