English
NVIDIA and over 35 founding partners have launched the Open Secure AI Alliance to build open-source AI tools, models, agent frameworks and security technologies. Its membership draws attention as much from absentees as participants: Microsoft, IBM, Cisco and Palantir are on board, while leading closed frontier model developers OpenAI, Google and Anthropic have not joined.
Supported by the Linux Foundation’s Akrites initiative and OpenSSF, the alliance aims to equip security teams with inspectable, adaptable technologies to protect AI models, agents and their software supply chains. Founding members cover cloud infrastructure, cybersecurity, enterprise software, AI research and open-source communities, including Adobe, Box, Capital One, Cisco, Cloudflare, CrowdStrike, Databricks, Dell Technologies, Elastic, Fortinet, GitHub, HPE, Hugging Face, IBM, LangChain, the Linux Foundation, Microsoft, Mistral, NAVER, NetApp, Nokia, Palantir, Palo Alto Networks, Perplexity, Red Hat, Salesforce, SAP, ServiceNow, Siemens, SK Telecom, Snowflake, SpaceXAI, Synopsys, Uber, vLLM and Zscaler.
The alliance stresses AI security assessment should cover the full agent stack—not merely model weights. An AI agent encompasses models, orchestration layers, tool integrations, identity rules, permissions, isolation, safety guardrails, logging and evaluation workflows. Though open models carry misuse risks such as disabled safeguards or abuse for cyberattacks, the alliance argues closed models do not resolve such hazards and block defenders from auditing model behavior, verifying controls or running AI on-premises.
It advocates flexible adoption of both closed and open AI: closed models suit select cutting-edge use cases, whereas open alternatives deliver transparency, local control and customisation for defensive tasks. This matters greatly for regulated industries and sovereign entities that require direct oversight of sensitive data, model behaviour, security logs and incident response.
NVIDIA will donate open models, weights, datasets and agent harness research to the alliance, with its NOOA framework — hosted on GitHub — as an early contribution. NOOA enhances the safety and auditability of AI agent harnesses, streamlining testing, tracing, governance and evaluation when agents interact with tools, external data, code bases and corporate systems. Planned joint work spans agent identity solutions, workload isolation, secure model formats, model scanning and secure AI-augmented software development.
Founding members will build components of an open AI defence stack. HPE backs SPIFFE/SPIRE for zero-trust workload identity to authenticate AI agents via cryptography. Hugging Face’s Safetensors, hosted under the PyTorch Foundation, eliminates arbitrary code execution risks in model weight storage for safer open-source model sharing. IBM and Red Hat promote Lightwell, leveraging signed patches to safeguard open-source supply chain integrity. Microsoft offers MDASH, a multi-model agent scanner detecting exploitable software flaws. SpaceXAI open-sourced Grok Build, a terminal AI coding agent, and plans to release Grok model weights, aligning with the alliance’s focus on independent auditing and modification of security-focused AI tools.
On policy, the alliance urges regulators to frame open AI tools as cybersecurity infrastructure instead of universal risks. Blanket restrictions on advanced open models risk forcing organisations to rely on a small pool of proprietary vendors and weaken defensive testing capabilities for businesses and governments. It calls for collective investment in open datasets, evaluation platforms, attack simulators, red-teaming utilities, secure model distribution and agent security controls.
For enterprise infrastructure teams, the initiative highlights that AI deployments must be secured as interconnected systems integrating models, applications, identity management, storage, networks, observability and automated decision systems.
Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!