Howard Lutonic, a candidate for the Ministry of Commerce, is suspected of competing with American artificial intelligence (AI) companies using US manufacturing technology after the US -made AI startup DeepSeek has swept the Internet and the stock market. I blamed China.
“I think they are [China] Just care about yourself and try to hurt us. And we need to protect ourselves, “Lutonic said on Wednesday when he was asked how to deal with China's competition threats.
“We need to promote innovation and stop helping them. You know, you know, you know, you know And you know, you know, you know, you know, you know, you know, you know You know, you know, you know, you know the open platform -Meta open platform.
“If they intend to compete with us, let them compete, but stop using our tools to compete with us.”
Lutnick mentions META's decision in 2023, publishing some of the AI technology so that ordinary people can build their own chatbots. This movement was facing the scrutiny at the time because other AI companies chose to keep technology under rock and key.
Lutnick stated that if the Senate confirmed him, he was planning to be “very strong” for this message.
“I'm excited to supervise the screw [the Bureau of Industry and Security] He is excited to adjust the tariffs and BiS to improve strength when we say no, and the answer must be no, “he said.
Deepseek's new AI model R1 has gained a strong popularity on the weekends and raised a bigger question about the hundreds of dollars invested in the AI infrastructure by US -made AI and billions of dollars. Sector caused a big sale.
DeepSeek's app jumped to the top of Apple's App Store on the weekend and stayed in the top of the top free app for at least three days. Chat-GPT maker Openai was delayed on the second list.
Founded in May 2023, the company claims that it spent only $ 5.6 million to train the latest models, but accessed chips in the Biden administration export management on semiconductor equipment to China. Is restricted.
Last year, Deepseek claimed that he had used a 2,000 -floor NVIDIA chip to train model V3 and R1, but some AI business leaders have questioned the company's claims.
LUTNICK expressed doubts about what was in the R1 model, and told the Senate Commercial, Science, and Transportation Committee that “I don't believe that DeepSeek was on the board.”
“It's nonsense,” he said.
When asked about Lutnick's claim, NVIDIA's spokeswoman said that the company was ready to cooperate with the government to pursue a unique approach to AI.
“The values set by the Biden administration are based on the performance level five years ago, and have been achieved by major games and workstations,” said Spokesman.
Earlier this week, NVIDIA's spokesman told Hill that Chinese companies are “excellent AI progress and perfect examples of test time scaling.”
“DeepSeek's research shows how to create a new model using the method, indicates that it is completely compliant with export control using a wide range of models.” Said Spokesman. “Progress requires a large number of NVIDIA GPUs and high -performance networking. Currently there are three scaling methods. Before training and after training, it is a new test time scaling.”
LUTNICK asked how the United States plans to continue investing, intellectual property, and technical break -throughs into enemy countries.
“Why is it more clear than this week, when the Chinese AI, Deepseek, was able to create things directly? How to use what they have taken from us. But it is ridiculous, and it is necessary to deal with it.
The hill arrived in deep -safe and meta for further comments.
