China Develops AI Fake News Detector Seen by Experts as Likely Censorship Tool

China Develops AI Fake News Detector Seen by Experts as Likely Censorship Tool
A woman works online in her cubicle at an office in Beijing on February 4, 2010. (FREDERIC J. BROWN/AFP/Getty Images)
4/10/2019
Updated:
4/10/2019

Chinese tech giant Alibaba officially announced April 1 that it has developed an artificial intelligence (AI) tool to combat the spread of fake news online.

Named the “Rumor Shredder,” the company claims that the tool is capable of verifying the authenticity of internet information, with an accuracy rate of 81 percent.

According to a March 1 report from Chinese media Qianjiang Evening News, the algorithm of this AI tool was developed by a group of researchers at Alibaba’s research and development lab for cutting-edge technology, DAMO Academy (Academy for Discovery, Adventure, Momentum, and Outlook).

Li Quanzhi, the chief scientist at DAMO who developed the project, told Qianjiang Evening News that the Rumor Shredder first traces the original post or the source of the information to make an initial judgment about the authenticity of the original publisher: whether it was a media outlet or an individual; what the publisher has circulated before; and whether the publisher has a record of spreading fake news.

It also will look at the trustworthiness of the platform where the news was first published. For example, the Rumor Shredder considers the state-run media Xinhua and other government platforms as “trustworthy” sources.

Then, the algorithm identifies key elements from the news article, and compares them to its database of already collected news articles—checking to see if the news appears logical and well-grounded. If the news appears to not be based on authoritative information, the authenticity score will be reduced.

This “AI rumor shredder” is expected to help reduce the workload for “social-media auditors,” according to Si Luo, chief scientist at DAMO Academy. Such auditors are hired by Chinese social-media platforms and the Chinese regime’s internet censors to monitor and screen out content that the regime deems inappropriate.

Li Yuanhua, formerly an assistant professor of history at China Capital Normal University who now resides in Sydney, told the Chinese-language Epoch Times in an April 5 interview that the AI tool will likely be used by the Chinese regime as a tool of suppression.

“Any pure scientific research, when in the hands of the Chinese communist regime, will be transformed into a tool to suppress dissident voices, and will eventually serve an evil purpose.”

Li pointed out that Alibaba researchers defined Xinhua as a credible website, when it is, in fact, part of the Chinese regime’s propaganda apparatus. “If a base of lies is defined as a trustworthy source, any further analysis is meaningless,” he said.

Independent China observer Gu He also believed checking for plagiarism is an excuse to mask the AI tool’s “internet police” function.

“This will help the Chinese government suppress the so-called ‘rumor spreaders,’” Gu said in an April 5 interview with the Chinese-language Epoch Times. The term is often used by authorities to charge dissidents critical of the Chinese regime.

“With the use of the ‘rumor shredder,’ the Chinese government can not only save manpower, but also crack down even harder on dissidents.”

Alibaba’s tool also drew the attention of Chinese netizens.

“Looking forward to using this product to evaluate China Central Television [state broadcaster], People’s Daily, and the Global Times [both state-run newspapers],” one netizen wrote in the comments section of a news article about the new tool.

Another netizen wrote: “The Chinese communist regime always claims that it is serving the Chinese people. Hope this AI can test if that is true.”

One netizen mentioned a taboo topic, the Tiananmen Square Massacre that occurred 30 years ago: “Can this AI tell us whether tanks crushed protesters dead during the 1989 Tiananmen student democracy movement?”