NVIDIA and Industry Leaders Unite in the 'Open Secure AI Alliance' to Focus on AI Safety

Most AI models are available for free or with limited free usage, including popular ones like ChatGPT. Paid subscriptions offer more tokens and features, but not everyone can afford them. Alternatively, open-source AI models exist, though they are fewer and the best often originate from China. NVIDIA is keenly interested in this space, as are dozens of U.S. companies that have joined the Open Secure AI Alliance, a coalition dedicated to developing open-source AI models.

NVIDIA recognizes that open-source AI models represent the present and future. Instead of relying on proprietary, token-based models that can become very expensive, there's a compelling alternative: AI models that run locally on available hardware and are free to use.

NVIDIA and Dozens of Companies Unite Under the Open Secure AI Alliance (OSAA) to Create Open-Source AI Tools

NVIDIA CEO Jensen Huang supports these models. Following the release of DeepSeek V4, he promptly added support for NVIDIA's Blackwell GPUs to achieve optimal performance. Now, NVIDIA and many other leading AI industry companies have formed the 'Open Secure AI Alliance' or OSAA. Notable members include Microsoft, IBM, Adobe, Dell, Hugging Face, and the Linux Foundation, all sharing a common goal: to develop open-source tools that improve AI and AI agents.

Open-source AI models can benefit various market segments, starting with cybersecurity. It's undesirable to use a closed-source AI model owned by a single company that controls all access and data. Through the Open Secure AI Alliance, they will follow the Linux Foundation's Akrites initiative, aiming to fix vulnerabilities using open, free, and accessible technology.

The Open Source AI Alliance Aims to Create Free, Open, and Accessible AI Models and Tools, with a Focus on Cybersecurity

With open source, all companies gain access to the latest technologies without discrimination, enabling them to protect themselves from cyberattacks. Notably, a recent, unexpected attack on Hugging Face highlighted this need; when existing solutions failed, they resorted to the Chinese open-source AI model GLM 5.2.

NVIDIA's CEO believes both open-source and closed-source models have their place. Closed-source models often excel in accuracy and attract greater development investment. While cybersecurity is a key focus, open-source AI models can also be employed for research, product development, and risk analysis and mitigation. The alliance also aims to focus on creating and controlling AI agents to prevent loss of control or the imposition of priorities by entities seeking more power and profit.