America's fear of China goes way beyond TikTok

It was a chilly night in Berkeley, California this past November when Sarah decided to stop by a buzzy after-party for an AI conference called The Curve. A year and a half ago, the 27-year-old had left her lucrative job as a trader in London to look for work in AI safety, which she considers the most important issue in the world, and she was eager to connect with others who felt the same. She certainly didn't expect to end the evening vilified as a Chinese spy.

At the party, the topic du jour was a recent article in The Economist, provocatively titled, "Is Xi Jinping an AI doomer?" Sarah, who was born in China a few hours from Shanghai, discussed the question with various AI researchers and policy analysts. Then one exchange turned sour.

"There's been rumors of espionage in Silicon Valley," Sarah recalled one person saying, "Like people preying on young, male, impressionable software engineers." The guy looked at Sarah. Uncomfortable, she excused herself from the conversation.

"I'm a Chinese national, but it's not like I'm a spy," she later recounted to another attendee.

Her comment was overheard by Samuel Hammond, an economist at the Foundation for American Innovation and an AI policy advisor for Project 2025. He posted it to X, where it attracted millions of views.

"So definitely a spy?" read one reply on X. "She was from Beijing and had a very posh accent," responded Hammond. (Sarah is from an entirely different part of China.) Other partygoers piled onto the speculation. "I'm glad I wasn't the only one who thought that," one said. Another insisted: "I can't dismiss the idea that she was CCP… If there's a plausible risk, & there is, she shouldn't be allowed in."

Sarah, who asked that I use just her first name, replied on X to clear up the confusion about why she left the conversation — it was cold and she had already been asked the same question by other people — but by then the exchange had reached escape velocity, spilling out across Silicon Valley. "Everyone knows," she said about the encounter. "I had to stop going to networking events. I just wanted this thing to die down."

As the race to develop advanced AI systems before China does ramps up, a new Red Scare has taken over the tech world. I spoke with Chinese workers and entrepreneurs in Silicon Valley, many of whom spoke on the condition of anonymity to avoid professional, personal, or legal repercussions, and found that Sarah's experience was far from rare. The same fear of Chinese espionage that forced TikTok to shut down is now pushing out Chinese-born AI professionals at the exact moment that American AI experts report a critical talent shortage in the field.

Concerns over Chinese spying have been on the rise since the early 2010s, when the US escalated its efforts to address Chinese cyberattacks. In 2018, the Department of Justice under Donald Trump undertook a controversial "China Initiative," which aimed to prevent industrial espionage in the research community by investigating hundreds of academics suspected of having ties to China. It found only a few cases of actual spying.

The plurality of investigations, Bloomberg and MIT Technology Review reported in 2021, involved undisclosed funding and affiliations to Chinese institutions, while "just three claim that secrets were handed over to Chinese agents," Bloomberg wrote. Instead of catching spies, many cases just indicted professors for bureaucratic oversights — those found guilty claimed they didn't disclose their funding because they didn't think they had to.

The China Initiative's slipshod approach upended many people's lives. Gang Chen, a nanotechnologist, and Anming Hu, a physicist, were both US-based tenured professors who were arrested by the FBI, only for authorities to later realize they had made a mistake. During the yearlong investigations, Chen was barred from his university's campus, and Hu was suspended without pay and his work authorization revoked. (Though both were later reinstated to their positions, Chen said he would no longer take federal research funding). The suicide of the neuroscientist Jane Ying Wu last year came after a China Initiative investigation shut down her lab.

The Biden administration escalated Trump's competitive approach toward China. The 2022 CHIPS Act, which aimed to bolster domestic production of semiconductors, the essential hardware powering AI, also restricted US investments, manufacturing, and research collaborations in China. Last April, Biden signed a bipartisan bill ordering the Chinese company ByteDance to sell TikTok on grounds of national security.

Now, the fear of espionage has shifted to Silicon Valley. In June last year, AI researcher Leopold Aschenbrenner said he was fired from OpenAI for sharing concerns about foreign espionage with the OpenAI board. He then published a paper-cum-manifesto titled "Situational Awareness: The Decade Ahead," about the race to AGI, or artificial general intelligence — what he calls the "most powerful weapon mankind has ever created," comparable to nuclear weapons. "If we're lucky, we'll be in an all-out race with the CCP; if we're unlucky, an all-out war," he wrote. In a section called "The Free World Must Prevail," Aschenbrenner paints a dystopian scenario where the Chinese Communist Party steals model weights and algorithms and uses them to target "advanced bioweapons" at "anybody but Han Chinese," "individually assess every citizen for dissent," and "enforce the Party's conception of 'truth.'" The paper was widely circulated in Silicon Valley and even shared to X by Ivanka Trump. (OpenAI has said the concerns Aschenbrenner shared with the board were not the reason he was fired.)

The venture capitalist Marc Andreessen entertained a similar thought experiment last March. "Let's assume that AI in 2024 is like atomic technology in 1943," he wrote on X. "What we see is the security equivalent of swiss cheese. My own assumption is that all such American AI labs are fully penetrated and that China is getting nightly downloads of all American AI research and code RIGHT NOW."

In response to mounting concerns, the Financial Times reported that Google, OpenAI, and several other US tech companies have tightened personnel screening. Some startups are turning to third-party tools like Strider Technologies, which scours public data to investigate individuals' connections to "state-sponsored risk," as Strider's website puts it. If a current or prospective employee is flagged by Strider's AI, FT reported that its "due diligence" process will then investigate their "family or financial links abroad as well as their travel history."

A spokesperson for Google said they hadn't stepped up their screening processes and told me that the company has "strict safeguards" for preventing the theft of trade secrets, "none of which are based on employees' nationality." OpenAI did not respond to multiple requests for comment.

Meanwhile, members of the Chinese diaspora in Silicon Valley are feeling the strain and paranoia of geopolitical tensions pressing on their social and professional lives. One 26-year-old woman who works at a San Francisco startup said she was once called a "honeypot" by a dance partner at a local bar. "He told me he worked on a nuclear-related company," she said. "He was like, 'I cannot share anything with you. Are you looking at my phone? You don't seem like a honeypot, but I have to be careful given that you're a Chinese national.'"

As a tech worker, she has turned down AI-related job opportunities to avoid scrutiny for herself and her family from both the US and Chinese governments. She also tries to downplay her Chinese accent and cultural identity. "A lot of Chinese nationals don't agree with the Chinese government" on issues like Uyghur repression in Xinjiang, she said. "But in social situations, you have to vocalize it passionately just to lower the guard people have toward you."

Sarah affirmed this sentiment. She hoped to contribute to global AI safety work by bridging the information gap between Chinese- and English-language researchers and policymakers, but the hawkish environment has made collaboration difficult. "Whenever you say anything neutral about China, people will think, 'That's pro-China, and this person is bought by the CCP,'" she said.

Many Chinese startup founders are also deemphasizing their nationality. One founder of a consumer AI startup told me that an investor asked her to remove the word "China" from a pitch deck and replace it with "Asia." It's a notable shift from the start of the century when top venture-capital firms were eager to invest in Chinese super-apps and open China-based funds.

HeyGen, a generative-AI startup founded in Shenzhen, dissolved its Chinese operation to rebrand and relocate to the US in 2023. It asked its Chinese investors to sell their equity, then raised a new Series A round from US- and UK-based funds to more easily purchase semiconductors under the CHIPS Act. One of HeyGen's original Chinese investors told me that it's now common for Chinese founders to turn down Chinese capital to avoid US governmental scrutiny.

Similarly, the TechCrunch journalist Rita Liao wrote that a Chinese company refused her coverage because her byline made them look "too Chinese." She said another Chinese founder told her: "We work for no government. We just want to build businesses."

Several people I spoke with described their situation as a "double bind." On one hand, they came to the US to pursue opportunities and liberties that Chinese society didn't afford them. But these days, eerily similar state and social sanctions are intruding on their work.

In some cases, the tensions are making it difficult for people from China to stay in the US. Ordinarily, foreign workers on visas who temporarily leave the US must get a stamp from a local US consulate before reentering. But more tech workers are being put through extra processing required for people working in sensitive industries before they can reenter the US, resulting in some people getting stranded abroad for months or years.

On forums for Chinese nationals, anxious students and professionals fret about how to avoid extended processing. "I work as a machine learning engineer in the Bay Area," reads one post from October. "I work on some AI product applications and do no core AI research. I thought that 'sensitive fields' referred to advanced defense research, but now I heard that any STEM field can be subject to checks. I'm feeling panicked."

Unlucky visa holders who get flagged have no choice but to appeal for an extended leave of absence from their university or workplace. One AI Ph.D. student wrote in November, "I have now waited for more than 130 days. My school has deferred me for one semester, but I cannot defer it again."

As tensions ramp up, many Chinese tech workers are reconsidering whether the American dream is worth the risk. When Microsoft offered its China-based employees a chance to relocate to another country, tensions between the US and China made some reluctant to take the offer, according to reporting by Rest of World. Going viral on X turned out to be a tipping point for Sarah. She returned to China in December. "Initially I was afraid of trouble on the Chinese side, but it turns out that the other side is more problematic," she said about the US.

Meanwhile, the US faces a critical talent shortage in AI. A 2021 report by the National Security Commission on AI found that the number of US-born STEM and AI doctorates is not nearly keeping pace with the industry's growth. While 42% of top-tier AI researchers worked in the US in 2022, only 18% of them received their undergraduate degree here, Macro Polo's AI Talent Tracker found. China is currently the largest source of these top-tier researchers — and more are choosing to stay in China.

"For the first time in our lifetime, the United States risks losing the competition for talent on the scientific frontiers," the NSCAI report says. "Immigration reform is a national security imperative."

Divyansh Kaushik, a tech and national security expert at Beacon Global Strategies, told me that America needs policies that are a "scalpel, not a sledgehammer." He recognizes the risk of espionage, citing China's 2017 National Intelligence Law, which legally obligates all Chinese citizens to "support, assist, and cooperate with state intelligence work." But Kaushik considers blanket bans on foreign nationals just as counterproductive. He instead pointed to policies that restrict students from specific Chinese-military-affiliated universities from obtaining visas. "There will be false positives and false negatives," but the US can mitigate the risk without overreaching, he told me.

Other experts believe that nationality-based anti-espionage efforts are more security theater than reality. Yangyang Cheng, a physicist and China researcher at Yale Law School, told me that AI risks "are not exclusive to Chinese firms or unique products of the Chinese authoritarian system." She cited examples of American professors who helped build biometric technologies for ethnic oppression in Xinjiang, arguing that we should be focused on preventing harm wherever it originates. She thinks the TikTok ban makes the same mistakes. "The focus on the Chinese government's subpoena power overlooks the many ways American companies cooperate with the state," she wrote for Wired.

It's unclear what stance this Trump administration will take. Some in Silicon Valley are hopeful that the president-elect will expand the visa program for high-skilled immigrants. During a recent intra-Republican fight over H-1B visas, Trump aligned with Elon Musk, telling supporters: "We need smart people coming into our country." But Trump's 2016 term oversaw higher costs, longer wait times, and increased denial rates for H-1B applicants. During a private dinner conversation in 2018 reported by Politico, Trump said, referring to China, "almost every student that comes over to this country is a spy."

Those who are staying in the US, meanwhile, say they feel exhausted. "I've been in the US for almost a decade," a Chinese-born data scientist and UC Berkeley graduate told me. "Many of us left to escape that political environment, and are the most liberal-leaning Chinese you can find. We spend so much time going through the American education and immigration system — and now the US says it doesn't welcome us either. It feels like we have this talent that no one wants."

Jasmine Sun is a writer covering tech, politics, and culture from San Francisco. She publishes a weekly newsletter on the "anthropology of disruption."