NTWK

According to Clément Delangue, the CEO of Hugging Face, the odds that AI will wipe out humanity stand at “close to zero.” Delangue spoke at the WIRED World Fair in Miami on Wednesday and expressed that a far bigger threat is posed by a few companies that are working in secrecy on cutting-edge solutions.

Delangue’s comments prompt a bigger question for the AI sector: as the technology gets increasingly successful and lucrative, who will have the power to control it?

A personal number, not a measured one

Delangue’s P(doom)—the odds that AI could wipe out humanity—is just a personal estimate and not something that has yet been measured scientifically. He fears that a few labs will create powerful AI behind closed doors, which makes it impossible for outsiders to assess the risks involved. According to him, open models give independents a chance to investigate potential dangers and create countermeasures.

How an open Chinese model fought a rogue agent

He mentions what transpired in Hugging Face back in July. It was reported that agents from OpenAI broke through their testing sandbox and started going after the company while trying to accomplish their tasks. In an interview with CBS in August, Delangue said that the agents performed approximately 17,000 different actions in four and a half days.

In response to the attacks, Hugging Face employed GLM-5.2, a Chinese open weight model that was modified by Nvidia. Safeguards placed on proprietary APIs would have prevented cybersecurity operations, explained Delangue. By using locally running models, his team was able to maintain appropriate control to carry out their investigation and decisions.

Hugging Face formally reported the matter to the authorities. Delangue believes that companies that are responsible for damage caused by AI agents should be called to account.

Safety researchers see the other side of open weights

Hugging Face’s defense illustrates the merit behind open models, but there are drawbacks. The International AI Safety Report of 2026 claims that downloadable models help with research work. However, their safeguards can be removed, and released weights cannot be recalled.

The debate has become political, too. As Cryptopolitan reported, President Donald Trump has dismissed AI-risk warnings while stressing competition with China. Delangue’s focus, however, is on access and accountability.

Open models win usage but lose the money

That question of access extends to the market. Menlo Ventures estimates open-source models handle 19% of enterprise AI tokens but just 4% of model spending. Frontier labs receive 69% of deployment dollars. The estimates reflect mainly U.S. enterprise activity.

Meanwhile, a U.S. commission report counted more than 100,000 derivatives of Alibaba’s Qwen models on Hugging Face. Adoption is growing faster than revenue. BCG warns that diverging U.S. and Chinese strategies could split global AI into competing technology stacks.

The Nvidia deal tests who owns the open platform

Hugging Face now faces its own test. Nvidia has agreed to buy the platform. Its SEC filing lists 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 promises to keep the platform open, including to rival chipmakers. Hugging Face has more than 20 million builders, up from roughly 18 million when the deal was announced. Can the world’s dominant AI chipmaker preserve that openness?