A COALITION of technology companies, cybersecurity firms and open source organizations has launched the Open Secure AI Alliance, saying open artificial intelligence (AI) technologies are essential to strengthening cyber defenses and improving transparency in AI security.
The alliance said open source software already underpins cloud computing, financial services, manufacturing, telecommunications, government operations and internet services. It argued that open AI models and tools should play a similar role in cybersecurity by allowing organizations to inspect, adapt and deploy AI systems within their own environments.
According to the alliance, cybersecurity requires both open and closed AI models, but open systems provide defenders with greater transparency, flexibility and control while reducing dependence on a single technology provider.
The group cited a recent security incident involving Hugging Face, saying the company relied on an open-weight AI model running on its own infrastructure to analyze more than 17,000 actions during a cyberattack after closed AI tools were unable to support its forensic investigation. It said the incident demonstrated the importance of allowing organizations to operate advanced AI systems on their own infrastructure during security incidents.
The Open Secure AI Alliance said it aims to develop and share open technologies, techniques and tools to help secure software systems and AI agents.
Founding members include organizations from the cloud computing, cybersecurity, enterprise software and AI sectors, among them Nvidia, Adobe, Cisco, Cloudflare, CrowdStrike, Dell Technologies, Fortinet, GitHub, Hugging Face, Hewlett Packard Enterprise, IBM, Microsoft, Mozilla, Nokia, Okta, Palo Alto Networks, Red Hat, Salesforce, SAP, ServiceNow, Siemens, Snowflake, VMware by Broadcom, Workday, Zscaler and the Linux Foundation.
The alliance acknowledged concerns that open AI models could be modified or misused for cyberattacks but argued that similar risks also exist with proprietary AI systems. It said effective safeguards should include rigorous testing, security evaluations, rapid vulnerability remediation and measures to prevent malicious use.
Several members also announced contributions to open AI security projects.
Nvidia said it will contribute open models, model weights, datasets and research related to AI agent harnesses. The company also released the open source Nvidia Labs Object-Oriented Agent (NOOA) framework on GitHub to support the testing, auditing and governance of AI agents.
Other contributions include Hewlett Packard Enterprise's work on the SPIFFE and SPIRE zero-trust identity framework, Hugging Face's Safetensors format for securely storing AI model weights, IBM and Red Hat's Lightwell project for digitally signed software patches, Microsoft's MDASH multi-model agentic scanning harness and SpaceXAI's open source Grok Build coding agent.
The alliance urged policymakers to recognize open AI models, security tools and evaluation frameworks as defensive technologies rather than security risks. It also called for greater investment in shared AI defense infrastructure, including datasets, evaluation frameworks, attack simulators and red-teaming tools, to support the secure development and deployment of AI systems.
