Natural language processing
MIT has terminated a research collaboration with iFlytek, a Chinese artificial intelligence company accused of supplying technology for surveilling Muslims in the northwestern province of Xinjiang.
The university canceled the relationship in February after reviewing an upcoming project under tightened guidelines governing funding from companies in China, Russia, and Saudi Arabia. MIT has not said why it terminated the iFlytek collaboration or disclosed details about the project that prompted the review, but it has faced pushback from some students and staff about the arrangement since it began two years ago.
“We take very seriously concerns about national security and economic security threats from China and other countries, and human rights issues,” says Maria Zuber, vice president of research at MIT.
US companies and universities have built ties with Chinese tech firms in recent years. But the relationships have come under increasing scrutiny as relations between the two countries have soured.
MIT announced what was supposed to be a five-year collaboration with iFlytek with fanfare in June 2018. Since then, iFlytek has helped fund a range of research on subjects including human-computer interaction, new approaches to machine learning, and applied voice recognition. Under the agreement, iFlytek selected existing projects to fund but MIT says the company did not receive special access to the work or receive proprietary data or code. The amount of money involved was not disclosed.
The arrangement became more controversial in October 2019, when the US government banned six Chinese AI companies, including iFlytek, from doing business with American firms for reportedly supplying technology used to oppress minority Uighurs in Xinjiang. In 2017, Human Rights Watch claimed iFlytek supplied police departments in Xinjiang with technology for identifying people using their voiceprints. Press reports paint a grim picture of widespread surveillance in the province, including the detention and disappearance of more than 1 million people.
iFlytek is one of China’s older AI companies, and while it specializes in voice recognition, it also offers tools for analyzing legal documents and medical imagery. Like other growing Chinese AI companies, contracts to supply software for processing video and audio to police departments and local governments are an important source of revenue.
The company said MIT’s decision was disappointing. “We are particularly sorry about this,” says Jiang Tao, a senior VP at iFlytek. “The vision of the cooperation was to build a better world with artificial intelligence together.”
Like other US universities, MIT receives funding from companies and individual donors, but several of its arrangements have proved controversial. In February 2019, the university reexamined funding from Saudi Arabia following the assassination of the journalist Jamal Khashoggi. The tighter guidelines for working with foreign companies were issued in April 2019 amid scrutiny of MIT’s relationship with two other Chinese companies, Huawei and ZTE. MIT had cut funding relationships with those companies in 2018 as the US government investigated their roles in alleged violations of US sanctions. In January 2020, MIT released the results of an investigation into funding from the convicted sex offender Jeffrey Epstein.
In 2018, MIT received a onetime donation of an undisclosed sum from SenseTime, another Chinese AI company now subject to the US government restrictions. The gift was reviewed by MIT’s Interim Gift Acceptance Committee, and an MIT spokesperson says there are no plans to return it.
US officials are increasingly wary of Chinese companies developing advanced technologies, amid rising trade tensions, accusations of intellectual property theft, and a heightened sense of international competition. Over the past two years, US intelligence agencies have repeatedly warned universities to watch for signs of espionage by Chinese students and professors, and prosecuted both Chinese-born and US academics for stealing intellectual property. In a meeting with senior figures at MIT in November 2019, Michael Kratsios, the US chief technology officer, warned against working with Chinese AI companies, according to a person familiar with the discussion.
Paul Triolo, a practice head at Eurasia Group specializing in global technology policy, says concerns over human rights violations are legitimate but the signals coming from the US government have been ambiguous. “Is this some sort of just punishment or really legitimate effort to try to change behavior?” he asks. “The danger is sort of painting them all with one brush, and not looking at what they're actually doing in Xinjiang, and how much they are taking steps to step away from that.”
Triolo says a complete unraveling of relations between the US and China will harm American AI too. He notes that China’s tech industry is making rapid progress in medical uses of AI, for example: “The flow of knowledge is not one way.”
MIT’s Zuber says the university doesn’t want to walk away from China. “We want to be able to draw the best talent in the world, and some of that best talent comes from China,” she adds. “The wrong thing to do is say we’re never going to work with these international entities under any circumstances and we’re just going to lock our doors.” Zuber also says “global collaborations are extremely important.”
When it comes to China, it may be difficult to ignore outcry over human rights issues. Zulkayda Mamat, a graduate student of Uighur descent who was critical of MIT’s ties to Chinese AI companies while studying there, welcomed the news but says MIT should scrutinize collaborations carefully. “I hope that it continues the process of reevaluation for all projects,” she pointed out. “[A] lack of vigilance will certainly put it on the wrong side of history.”