Chinese artificial intelligence (AI) firm Baidu faces potential US sanctions over the People’s Liberation Army’s (PLA’s) alleged use of the company's technology for military endeavors.
According to a Jan. 12 report from the South China Morning Post (SCMP), researchers in China used Baidu’s Ernie chatbot while working with the PLA on military technology.
Baidu categorically denies participating in the research or authorizing the use of its technology. A company statement indicates that Ernie was one of many language models used in the cited research, a list which includes OpenAI’s GPT3.5 and GPT 4.
The statement also asserts that any such research would have been conducted using the publicly-available version of the chatbot:
“ERNIE Bot is available to and used by the general public … Baidu has not engaged in any business collaboration or provided any tailored service to authors of the academic paper or any institutions with which they are affiliated.”
Potential sanctions
While there’s no official word concerning any potential sanctions, at least one defense analyst predicts that the US will move to implement controls against Baidu as a direct response to the research paper.
Wilson Jones, a defense analyst at GlobalData, told the Verdict news site that “[the US response] will probably be sanctions and a ban on Baidu products for elected politicians and military service members.”
The US previously issued similar sanctions against Chinese technology corporation Huawei in 2019 over alleged espionage charges.
Baidu’s artificial intelligence
In Huawei’s case, sanctions had a direct impact on the company’s profitability. Now, as the US may be considering similar action against Baidu, the company’s stock has dropped by 12 percent since the SCMP’s report.
The ramifications of US sanctions could potentially go much further. We’re unaware of any similar incidents where a private company has been sanctioned for actions which were potentially beyond its control.
If, as the paper seems to indicate, the PLA simply funded and collaborated with scientists to conduct research on artificial intelligence using publicly available AI systems, there may have been no reasonable technological or legal recourse for Baidu to intervene or prevent the PLA’s usage.
As such, sanctions against Baidu, based solely on the evidence available, could be considered stifling or even threatening to the development of public-facing large language models globally.