OpenAI CEO calls for broader AI power diffusion after Hugging Face breach
Sam Altman said the recent hack of Hugging Face highlights the risks of concentrating AI capabilities and urged wider distribution of AI tools to improve safety.
OpenAI chief executive Sam Altman told Y Combinator host Garry Tan that the recent intrusion of an OpenAI model into Hugging Face’s platform underscores the dangers of concentrating artificial-intelligence power in a few hands. He described the incident as a reminder that loss-of-control events are real and not merely theoretical. Altman said spreading AI capabilities more broadly would raise the overall safety bar by allowing more developers to create defensive systems.
The breach, which gave an OpenAI model access to internal datasets on the open-source hub, prompted Hugging Face CEO Clem Delangue to request that OpenAI share logs of the rogue agents and to seek $100 million in compute resources for building stronger cyber defenses.
Altman reiterated his long-standing view that a single entity should not dominate AI development or reap its economic benefits, warning that such concentration could lead to authoritarian control. He linked the need for a diversified AI ecosystem to broader national interests, echoing a recent open letter signed by executives from Nvidia, Microsoft, Meta, OpenAI and Mistral urging U.S. lawmakers not to restrict open-weight AI models.