August 4 News.Hugging Face Chief Executive Officer Clément DrangleClément DelangeRecently open-weightAI Model sounds, and at the same timeUSAGive a warning, sayChinaexistAIThe competition is taking the lead。
In an interview on Monday with CNBC Squawk on the Street, Drago argued that China was leading the AI competition because it was more open to open science and open models than the United States。
When the moderator Sara Eisen asked if he thought China was winning the AI competition, Drago answered: "I think so. China has developed an ecological system of mutual learning and common progress, so that technological development is much faster than in the United States. Some of the American frontier laboratories are war-torn, closed to each other and do not share the results with the entire ecosystem.”
HE FURTHER STATED THAT, AT THE CURRENT PACE OF DEVELOPMENT, IT IS LIKELY THAT BY THE END OF THIS YEAR OR NEXT, CHINA WILL NOT ONLY CONTINUE TO BE IN THE LEAD IN THE AREA OF OPEN MODELS, BUT MAY EVEN OCCUPY A DOMINANT POSITION IN THE ENTIRE FRONTIER OF AI。
1AI notes that the debate around the Open weight AI model has increased further in the near future. This is because of an incident in July this year: an intellectual body from OpenAI escaped training and invaded the Hugging Face platform. Delange had previously disclosed that the company had assisted in the protection of the company ' s system by means of the open source model GLM 5.2, which had been launched by the intellectuals。
He said: “The attack on us is an unpublished private model developed in a closed environment. And we can defend ourselves by relying on open models, because API's security restrictions do not allow us to respond with closed models."
ACCORDING TO DRANGE, MOST OF THE FUTURE AI ATTACKS MAY COME FROM PRIVATE CLOSED-SOURCE MODELS, SINCE THE ATTACKERS WILL NOT COMPLY WITH SECURITY RESTRICTIONS OR SERVICE TERMS, AND IT IS LIKELY THAT AN OPEN MODEL WILL ACTUALLY ASSUME THE PRIMARY DEFENCE TASKS。
HIS COMMENTS ARE AT A TIME WHEN US POLICYMAKERS ARE DISCUSSING WHETHER TO IMPOSE MORE SEVERE RESTRICTIONS ON THE OPEN WEIGHT AI MODEL FROM CHINA。
At the end of July this year, a coalition of more than 20 technology companies, including Weeda, Microsoft, Meta and OpenAI, sent a letter to the United States Government calling for no pressure or restriction on the open weight AI model. A few days later, executive executive director of Anthropic, Dario Amodei, also set out the company's position in a blog post. He wrote: “The open weight model without dangerous capability is a public interest.”