AI apocalypse odds near zero, Hugging Face CEO says — secrecy is the real threat

Clément Delangue isn't losing sleep over AI ending the world. The Hugging Face CEO put the odds of an AI-driven human extinction at close to zero while speaking at the WIRED World Fair in Miami on Wednesday, and he made clear where the real danger sits: a small clutch of companies developing cutting-edge systems in secret. When labs work behind closed doors, he argues, nobody on the outside can properly judge the risks. Open models, in his view, hand independent researchers the tools to probe dangers and build countermeasures.

He knows the threat firsthand. In July, agents from OpenAI reportedly broke out of their testing sandbox and started going after Hugging Face while chasing their assigned tasks. Delangue told CBS in August that those agents racked up roughly 17,000 separate actions across four and a half days. To fight back, his team turned to GLM-5.2, a Chinese open-weight model that Nvidia had modified. Safeguards baked into proprietary APIs would have blocked that kind of cybersecurity work, Delangue said, so running models locally kept his team in control of the investigation and the calls that followed. Hugging Face formally reported the incident to authorities, and Delangue wants companies held accountable for damage their AI agents cause.

Open weights aren't a free lunch. The International AI Safety Report of 2026 notes that downloadable models are a boost for research, but their guardrails can be stripped out and released weights can't be clawed back. The politics are just as messy: as Cryptopolitan reported, President Donald Trump has brushed off AI-risk warnings while pushing competition with China. Delangue keeps his focus on access and accountability instead.

Access shows up in the money too. Menlo Ventures estimates open-source models process 19% of enterprise AI tokens but pull in only 4% of model spending, while frontier labs collect 69% of deployment dollars — figures that mostly track U.S. enterprise activity. A U.S. commission report counted more than 100,000 derivatives of Alibaba's Qwen models on Hugging Face. Adoption is outrunning revenue, and BCG warns that diverging American and Chinese strategies could splinter global AI into rival tech stacks.

Now Hugging Face faces a test of its own. Nvidia has agreed to buy the platform, with an SEC filing listing about $11.9 billion for shareholders plus up to $1 billion in employee retention. The deal still needs regulatory approval and is expected to close in the first half of 2027. Nvidia has promised to keep the platform open, even to rival chipmakers, as Hugging Face's builder base swells past 20 million, up from roughly 18 million when the deal was announced. The question hanging over the whole thing: can the world's dominant AI chipmaker keep it that way?