- In a letter to the US government, Openai slammed her rapid Chinese rival Deepseek.
- “While America holds a lead in him today, Deepseek shows that our lead is not wide and is narrowing,” Openai wrote.
- The letter also described policy recommendations to ensure America’s leadership.
Openai has launched an attack on his rapid Chinese rival, Deepseek.
In a 15-page letter to the US government on Thursday, Openai called the latest model of Deepseek, R1, a “important” development-which signals the growing ambitions of China and underlines the coercive competition between rival countries.
“While America holds a lead in him today, Deepseek shows that our bullet is not wide and is narrowing,” wrote Chris Lehane, the Vice President of Openai on Global Affairs, wrote to the Office of Science and Technology Policy.
Founded in 2023 by Chinese entrepreneur Liang Wenfeng, Deepseek shocked the industry of him and the US market with his low -cost reasoning model R1, discovered in January. The company said its R1 model rivals main competitors, such as O1 of Chatgpt, but with a cost.
The company after Chatgt did not keep its criticism of the Chinese model again. Lehane warned that the use of Deepseek in critical infrastructure and other high -risk applications is a “significant risk” because Deepseek could be pressure from the Chinese government to manipulate its models.
The US Navy has announced members of the use of Deepseek’s use, and Taiwan banned it from government agencies in February on security concerns.
“Because Deepseek is at the same time subsidized by the state, controlled by the state and freely available, the cost to its users is their intimacy and security,” Lehane wrote.
Lehane also said that Deepseek is “more ready” to generate responses to illegal and harmful activities, including identity fraud and intellectual property theft.
“The Chinese Communist Party looks at violations of American intellectual property rights as a feature, not a flaw,” he wrote.
Lehane warned that China would use it as a geopolitical tool offering Deepseek to countries that need it and infrastructure funding – an action associated with its existing belt and road initiative. For more than a decade, China has used this program to spend more than $ 1 trillion on infrastructure programs globally.
Openai’s general manager, Sam Altman, called R1 to Deepseek a “invigorating” competitor in January, saying it will push his company to release “better models” faster.
Providing America’s leadership in it
Openai’s letter also described policy recommendations to ensure America’s leadership in him.
The letter was written in response to the request of the White House science and technology policy for public data in February.
In January, President Donald Trump signed an executive order demanding that the US “maintain and improve the global dominance of America” ​​as they revoke the Biden administration order demanding greater transparency from his companies.
His administration now has a duty to develop an action plan until July.
Lehin on Thursday proposed a series of “freedom -focused” policies, including a regulatory strategy that would facilitate developers and the American ones not to enforce the “extremely heavy state laws”.
The company also said that the US government can facilitate companies and it will be trained in copyright protected materials, arguing that strict copyright laws can slow down innovation and restrict access to training data.
Openai and her peers have been criticized for using copyright protected content to train models. The authors and newspapers, including the New York Times, have sued Openai, saying the beginning is violating the copyright law.
Lehane proposed a copyright strategy that would protect creators’ rights while defending the leadership of America and national security.
“The federal government can ensure the freedom of Americans to learn from him, and avoid the loss of our lead to the PRC while maintaining the ability of American models to learn from copyright protected material,” he said.