Hugging Face CEO: AI risks come from closed labs, not open-source systems

Hugging Face CEO Clément Delangue told Wired that his assessment of the probability of AI causing existential harm is "close to zero," positioning open-source AI development as safer than closed alternatives.

Delangue argued that the concentration of risk lies with a small number of companies building AI systems behind closed doors, not with publicly available open-source models. He framed open-source development as more transparent and distributed, implying greater safety through visibility and community oversight.

This stance reflects Hugging Face's business model and philosophy, which centers on democratizing access to AI tools and models. The debate over open versus closed AI development has become central to discussions about AI regulation, safety, and competitive advantage.