10 August 2017

Can the US really stop China’s access to American AI tech?

By DOUG TSURUOKA

Washington is buzzing with reports that the US plans to heighten scrutiny of Chinese efforts to access artificial intelligence (AI) and other sensitive technology by acquiring or taking stakes in cutting-edge Silicon Valley firms.

The worry is that China, or others, will use acquired American AI tech to ramp up their military and economic power. The Pentagon is said to have a report under wraps that says China is dodging US security reviews by engaging in outwardly harmless joint ventures, investing in early-stage startups or gaining minority shareholdings in US tech companies.

But a larger question is whether it’s practical or even possible for the Committee on Foreign Investment in the United States (CFIUS) – the inter-agency committee that reviews foreign acquisition of US firms on national security grounds – to restrict Chinese access to AI, especially given that AI is on the verge of becoming an integral part of many mainstream technologies. Another issue is whether Beijing, which already invests heavily in the sector, is really making a concerted effort to steal stateside AI tech while skirting US oversight.

Former US officials and defense analysts say that Washington has no choice but to act. They note precedents set by past US efforts to control outflows of sensitive technology. Their hope is that this can be done in the case of AI, though signs are that the technology may become too prevalent to effectively police.

“Anyone concerned about US national security would have to be concerned about the Chinese making off with crown jewels such as AI,” John Pike, an intelligence expert who heads the military think tank GlobalSecurity.org told Asia Times. 

John Schaus, an Asian security expert at the Center for Strategic & International Studies (CSIS) in Washington, says the US must look harder at so-called “backdoor” investments by Chinese firms in Silicon Valley. While bids by Chinese firms with links to China’s military can be blocked, he says it’s tougher to pinpoint with more conventional companies how much autonomy they have and if China’s government can access their technology.

Schaus says US concerns about national security are justified since AI is a new field whose military and economic implications are still being calibrated. “We’re really at the front end of understanding AI,” said Schaus, a former Pentagon official whose duties included day-to-day management of the US-China military relationship.

He points to the use of export controls during the Cold War to halt the shipment of sensitive tech to the Soviet Union as a successful example of such policing. Current US regulatory regimes include International Traffic in Arms Regulations (ITAR) that control the export of military-related technologies. The US also has cryptography technology export rules in place.

Schaus, at the same time, concedes it’s problematic whether the US can effectively curb Chinese access to AI. “It’s a good question,” Schaus said, acknowledging the difficulty of controlling a spreading technology.

Global Security’s Pike says the practicality of such moves depends on exactly what the US plans to do as part of a strategy to limit Chinese access to AI. But one thing’s for sure, Pike says: “The Chinese are not our friends. They have a vision of an international order that is quite different from the US.”

Winn Schwartau, a noted cyber warfare expert, says it would be a serious mistake to curb Chinese investment in US AI firms. “If the US tries to crack down, that’s the complete Balkanization of intellectual property. It’s a failed fortress-mentality mindset,” he told Asia Times. “AI is a global science and this is a global society.”

In weighing the AI security issue, Schaus says the potential threat should be divided into two parts: foreign access to US R&D in AI and foreign access to actual AI expertise and products. He says the US must eye both the long-term and immediate challenges posed by such access. 

Is the Chinese government making a coordinated effort to access US AI tech? Schaus says the evidence is spotty, though he asserts there have been cases of Chinese firms with ties to China’s government attempting such transactions.

“If the US tries to crack down, that’s the complete Balkanization of intellectual property. It’s a failed fortress-mentality mindset. AI is a global science and this is a global society”

He says this was the case with a US$1.3 billion bid in November by a China-backed buyout fund, Canyon Bridge Capital Partners, to acquire the Portland, Oregon-based Lattice Semiconductor Corp. 

Reuters reported last year that Canyon Bridge is funded by cash originating from China’s central government, and also has indirect links to its space program.

Researcher CB Insights says 29 investors from mainland China have invested in US AI-focused companies since 2012. Chinese investors, the researcher says, invested a total of US$9.9 billion in US tech startups in 2015. The level was four times higher than the previous year.

Does China already lead in AI?

Other developments stir doubts about how crucial accessing US AI technology is to China’s overall AI campaign.

A July 15 article in the Economist noted that “China May Match or Beat America in AI.” The story noted that China may already be overtaking the US based on its “deep pool” of data in the field and recruitment of world-class AI experts. 

An Obama administration report in October 2016 said that China had overtaken the US in publishing scientific articles on deep learning — a key branch of AI.

“China has it’s (stuff) together,” said Schwartau, who is chairman of Mobile Active Defense, a US firm that specializes in security and compliance for smartphones and tablets.

China has made no secret of its plans to surpass the US in AI. Its State Council issued a development blueprint on July 20 calling on the country to become the world’s leader in AI by 2030 with a domestic industry worth US$150 billion.

No comments: