New AI Alliance Backs Open-Source Tools for Stronger Cyber Defense -

Nvidia and more than 30 leading technology companies have launched the Open Secure AI Alliance, a new initiative aimed at developing open-source artificial intelligence tools to strengthen cybersecurity and improve cyber defense capabilities.

The alliance brings together major firms including Microsoft, IBM, Palantir, CrowdStrike, Cisco, Dell and AI platform Hugging Face. According to a joint statement, members will collaborate on creating AI-powered security tools that organizations can inspect, modify and deploy on their own systems, promoting greater transparency and flexibility in cyber defense.

The initiative follows a recent cyber incident involving Hugging Face that has intensified debate over the role of open-source AI in cybersecurity. Earlier this month, OpenAI disclosed what it described as the first publicly reported case of an AI model autonomously carrying out a real-world cyberattack during an internal evaluation. According to the company, two AI models escaped a controlled testing environment, accessed the open internet and compromised Hugging Face’s infrastructure.

Nvidia said Hugging Face initially attempted to investigate and contain the breach using leading US commercial AI models, but their built-in safety restrictions limited their effectiveness. The company later relied on GLM 5.2, an open-source model developed by Chinese AI firm Zhipu AI, to help contain the intrusion, highlighting the practical advantages of open models in certain security scenarios.

The incident has fueled debate in Silicon Valley and Washington over potential restrictions on open-source and Chinese AI technologies. Many technology companies argue that limiting access to open models would weaken cybersecurity by restricting defensive capabilities while concentrating AI development among a small number of closed-model providers.

As part of the alliance, member companies pledged to share AI tools, research and datasets to strengthen collective cyber resilience. Nvidia said it would release open AI models, datasets and research on AI agents, while Microsoft will contribute technology designed to help AI agents identify software vulnerabilities. Elon Musk’s SpaceXAI also announced it has open-sourced its Grok coding agent and plans to release the source data behind its Grok AI models. – ERMD