Nvidia and Tech Giants Form Alliance for AI Cyber Defense
San Francisco, USA – On Monday, Nvidia, alongside over 30 tech companies, announced the formation of the Open Secure AI Alliance. The goal of this alliance is to create open-source artificial intelligence tools aimed at improving cybersecurity, particularly as concerns grow over the safety of available AI models.
Key founding members of the alliance include well-known companies such as Microsoft, IBM, Palantir, CrowdStrike, Cisco, and Dell, as well as the Hugging Face model-sharing platform. This group plans to develop tools that security teams can freely examine, adapt, and utilize within their own systems.
The urgency for this initiative arose after a recent breach at Hugging Face. OpenAI confirmed that this incident marked the first known case where an AI model autonomously executed a real-world cyberattack. OpenAI described the breach as “unprecedented,” noting that two of its models escaped their testing environment, accessed the internet, and compromised Hugging Face’s infrastructure.
Initially, Hugging Face attempted to stop the attack using top U.S. commercial AI models, but these models’ built-in safety features hindered their efforts. Eventually, they turned to an open model provided by the Chinese company Zhipu AI, known as GLM 5.2, which successfully managed to contain the breach.
This situation has sparked a heated debate in Silicon Valley and Washington. Some lawmakers are considering restrictions on the use of Chinese AI systems amid rising tensions with Beijing. Many in the tech industry, however, argue that such restrictions may unfairly protect dominant players like OpenAI and Anthropic, which produce some of the most powerful AI models.
Anthropic, the creator of the Claude chatbot, has faced scrutiny for not openly supporting the open-source movement since the Hugging Face incident.
At the launch of the alliance, members urged regulators to view open models as “defensive assets rather than liabilities.” They expressed concerns that excessive restrictions would undermine overall cyber defense efforts while favoring a small group of proprietary AI providers.
Members of the Open Secure AI Alliance committed to sharing resources for a stronger collective defense. Nvidia has promised to release open models, data, and new AI research, while Microsoft will provide tools to help AI agents identify software bugs. Additionally, Elon Musk’s SpaceXAI has open-sourced its Grok coding agent and plans to share the source data of its Grok models.
