OpenAI Cyberattack Fallout Continues as Nvidia, SpaceX, and Microsoft Launch Open Source AI Security Alliance
NVIDIA, together with dozens of global technology giants, has officially launched an artificial intelligence (AI) security initiative focused on the development of open models. The move follows a cyber attack incident involving OpenAI’s AI models against the Hugging Face platform.
The alliance, named the Open Secure AI Alliance, includes major companies such as Microsoft, SpaceX and Palantir, along with dozens of other technology firms from the United States and Europe. Its primary focus is to fix and disclose system vulnerabilities using open technology.
Last week, it emerged that the start-up Hugging Face had been targeted by a cyber attack carried out by uncontrolled (rogue) OpenAI models. In the incident, Hugging Face was reportedly unable to use mainstream US models to defend itself because the safety guardrails on those closed models could not distinguish between attackers and defenders.
As a solution, Hugging Face turned to Chinese open-weight models hosted independently. These models proved effective because they were not bound by the same infrastructure restrictions as US closed models.
“The recent Hugging Face security incident provides a clear reminder: cyber defenders need open frontier agent systems for self-defence,” Nvidia wrote in its official statement.
The launch of the initiative comes amid political pressure in Washington to limit the adoption of Chinese AI models. The US government is concerned about the practice of “distillation”, whereby Chinese models extract knowledge from more advanced US models.
US Treasury Secretary Scott Bessent even threatened to impose sanctions on Chinese companies that carry out distillation attacks against American firms. Chris McGuire of the Council on Foreign Relations said the US government might impose restrictions on transactions or the use of APIs for those Chinese models.
However, the technology sector is pushing back. Nvidia, Microsoft, Meta and Palantir have released a joint letter urging policymakers to avoid “premature restrictions” on open-weight models.
Nvidia stressed that when cyber defence teams cannot inspect, adapt and run advanced AI on their own infrastructure, their ability to respond to attacks will be severely limited.
“The practical truth is that speed is decisive. Restrictions on open models are feared to hinder innovation and shift the centre of technological progress abroad,” the statement concluded. (CNBC/H-3)