• by Laurie Sullivan , Staff Writer @lauriesullivan, January 15, 2024

As reports surface that Chinese search engine Baidu's AI technology has been linked to key military research, ChatGPT maker OpenAI has changed the fine print in its use policy that eliminates text related specifically to the tech for military and warfare.

Baidu is often compared with Google’s strength in technology, from search to AI.

A report that surfaced Monday now links Baidu’s Ernie AI platform to People’s Liberation Army (PLA) scientists, the wing of the Chinese Communist Party (CCP), who are reportedly using the technology to train a military AI system that can better predict the behavior of human adversaries.

The allegations -- similar to what many U.S.-based technology experts have warned against -- raise questions on whether this technology could attract U.S. sanctions.

South China Morning Post -- owned by Alibaba, an online conglomerate of publishers and marketplaces in China -- reported that Baidu’s stock sank 12% to HK$100.50 after reporting that scientists at Chinese military labs were testing an artificial intelligence system based on the company’s LLM, Ernie.

Baidu on Monday denied knowledge of the project, but in a statement on its website acknowledged being aware of reports of the academic paper that mentions several of its large language models (LLMs) such as GPT3.5, GPT-3.5-turbo, GPT4, HTML-T5, and its Ernie Bot that competes with OpenAI’s ChatGPT and is available to the general public.

“The academic paper, published by scholars at a Chinese university, described how the authors built prompts and received responses from LLMs, using the functions available to any user interacting with generative AI tools,” Baidu wrote in a statement. “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.”

The statement went on to explain that Baidu is committed to operating its AI products and businesses in compliance with applicable laws, regulations and best corporate practices.

Hong Kong traders cited the South China Morning Post report, which detailed how a university affiliated with the PLA’s Strategic Support Force, which oversees cyberwarfare, had tested its AI system on Baidu’s ChatGPT-like Ernie, Bloomberg reported.

A link in the article was provided that led to the research paper, but now it leads to a dead page with a “404” error message.

Computerworld reports that OpenAI changes in the company’s use policy was done prior to January 10. It called out and disallowed the use of OpenAI models for weapons development, military and warfare, and content that promotes, encourages or depicts acts of self-harm.

The list has been condensed to describe Universal Policies that disallow anyone to use its service to harm others and bans the repurposing or distribution of output.