New Delhi, April 24 (IANS). Amidst the fear of online attacks and leakage of confidential information, America has decided to increase cooperation with its artificial intelligence companies. The Trump administration has taken this step after claims that China has been accused of trying to weaken American research and development and gain access to confidential information.
According to a BBC report, Michael Kratsios, director of the White House Office of Science and Technology Policy, said in an internal memo that new evidence has emerged claiming that foreign entities are abusing American companies through a process called distillation.
In fact, distillation in AI is a technique in which a large, complex model teaches its learned information to a smaller and faster model.
Michael Kratsios, science advisor to the President of the United States, says that this strategy of China is an attempt to weaken American research and development and gain access to confidential information.
To stop these activities, the White House will now share more information with American AI companies. Like how these attacks are carried out and who are the people involved in it. Along with this, better coordination will be created with the companies, so that these attacks can be prevented and best practices can be prepared to deal with them. Besides, ways to hold such foreign elements responsible will also be explored.
In a distillation campaign, companies create thousands of fake accounts and use an AI tool or chatbot to make them look like a normal user. These accounts then work together to “jailbreak” the AI model or extract information that should not be public. After this, that information is used in creating and training our AI models.
Kratsios said that as methods to detect and stop these large-scale distillations become better, questions will be raised about the reliability of such foreign models, which are built on a similarly weak foundation.
A representative of the Chinese Embassy in Washington DC rejected these allegations and said that China’s technological progress is the result of its own hard work and international cooperation.
US AI company Anthropic had alleged in March that three Chinese unicorn companies DeepSeek, MiniMax and Moonshot AI improved their systems by illegally extracting capabilities from its cloud model.
–IANS
AY/VC
