Artificial intelligence infrastructure is becoming increasingly complex, but the security knowledge needed to protect it remains distributed across hardware manufacturers, cloud providers, infrastructure operators, and independent researchers.
That fragmentation creates a challenge for an industry where different infrastructure components are closely interconnected. A vulnerability in server firmware, for example, may have implications for management systems, customer workloads, and the isolation between tenants. Yet organizations do not necessarily share a common framework for evaluating these risks.
A new initiative from Lava Security aims to help address that problem. The company announced that its Top 10 Neocloud and AI Data Center Security Risks framework has become an official OWASP project, bringing the research into a community-driven environment for continued development.
The framework was developed by Yakir Kadkoda, CTO and Co-Founder of Lava Security, and Michael Katchinskiy, Head of Security Research. Its initial development incorporated research and feedback from more than 20 security and infrastructure practitioners.
The Problem With Fragmented Infrastructure Security
AI data centers are not simply traditional data centers equipped with additional computing capacity. They combine thousands of GPUs, specialized management controllers, high-speed networking, shared storage, and complex orchestration systems.
Many of these technologies were originally designed for trusted high-performance computing environments. However, their increasing use in commercial neocloud environments introduces a different security model.
In these deployments, unrelated organizations may share physical infrastructure, while customers can receive extensive control over their rented servers.
That arrangement creates risks that conventional security practices may not fully account for. One example involves hardware boot configurations that can persist even after an operating system reinstall. If a tenant modifies these settings through an unrestricted management interface, the changes could remain in place when the server is reassigned to another customer.
The challenge is not limited to individual vulnerabilities. It also concerns the absence of a consistent way for different organizations to understand, communicate, and address infrastructure-level threats.
“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.”
What the OWASP Framework Covers
The OWASP Top 10 Neocloud and AI Data Center Security Risks takes a broad approach to infrastructure security, examining potential weaknesses across hardware, firmware, networking, management, and operational systems.
Its categories include compromised server firmware, vulnerabilities in InfiniBand and RDMA networking technologies, inadequate isolation between tenants, and insecure out-of-band management interfaces.
Other areas include supply chain threats, data center facility systems, shared storage and AI artifacts, incomplete security certifications, centralized operational services, and delays in vulnerability patching.
The framework is designed to connect technical weaknesses with their potential consequences. Each risk category includes attack scenarios, real-world implications, and practical recommendations for mitigation.
This structure gives infrastructure providers and operators a reference they can use to assess their environments and identify areas requiring attention. It also offers organizations purchasing GPU cloud capacity a way to better understand the security controls protecting their workloads.
The emphasis is on examining the entire infrastructure environment rather than evaluating individual technologies in isolation.
Why Community Participation Matters
The framework was first published in July 2026. Its development drew on disclosed vulnerabilities, incident reports, technical documentation, and Lava Security’s research into the security challenges affecting AI infrastructure.
More than 20 practitioners contributed feedback, representing neocloud providers, hardware OEMs, security companies, universities, and research institutions.
Moving the initiative into OWASP creates a structure through which that collaboration can continue. Instead of remaining a framework maintained solely by one company, it can evolve through contributions from different parts of the industry.
“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.”
Building a Common Reference for an Emerging Market
The project will continue evolving through an open development process on GitHub. Security researchers, infrastructure engineers, hardware vendors, and operators can review its content, challenge existing assumptions, suggest changes, and contribute additional research or mitigation recommendations.
Contributors will also receive credit in future releases.
The initiative’s development model reflects the changing nature of AI infrastructure itself. New GPU architectures, networking technologies, management systems, and deployment models can introduce additional security considerations, making an evolving reference more useful than a static assessment of existing threats.
For an ecosystem that spans infrastructure providers, hardware manufacturers, and organizations running AI workloads, a shared framework can help establish a common language around security expectations.
The OWASP project does not eliminate the technical challenges associated with securing AI infrastructure. It does, however, create an open starting point for bringing together the research, operational experience, and expertise needed to address them.
More information is available through the official OWASP project page and its GitHub repository.
John Kevin Hao is a news and feature writer covering cybersecurity, technology, and business targeted for professional audiences.