CrowdStrike, a prominent player in U.S. cybersecurity, has raised alarms about a surge in cyberattacks from Chinese entities, specifically aimed at pilfering artificial intelligence technologies to lessen the competitive edge of the United States. Matt Pearl, who heads the strategic technologies program at the Center for Strategic and International Studies, pointed out that, as the race for AI dominance intensifies, the focus from these actors is shifting from traditional tech-related hacks to human vulnerabilities within organizations. Pearl noted, "The People’s Republic of China is increasingly targeting the tech sector," highlighting that the hackers are no longer just interested in specific trade secrets, such as hardware blueprints, but are instead seeking any information that could help close the estimated three- to four-month gap in AI development with the U.S. This encompasses identifying a company's product strategies, particularly in fiercely competitive markets, as well as probing weaknesses within supply chains.
Evidence of these activities is mounting. In a report released in June, CrowdStrike revealed that entities linked to China represented more than half of state-sponsored intrusions into technology firms, particularly those involved in AI, during the year leading up to March 31.
In another development, American tech startup Anthropic has accused companies such as Alibaba of unlawful attempts to acquire its AI technology. Alibaba has not yet responded to requests for comments regarding these allegations. Last year, Copyleaks, a U.S.-based AI detection startup, reported that responses from the Chinese company DeepSeek's R1 model bore close resemblance—about 75%—to those produced by OpenAI's ChatGPT. This raises concerns that the open-source model from China might have used training data from the U.S. model. Copyleaks’ CEO Alon Yamin noted, "We've not seen a similar stylistic match in other large language models."
Additionally, Brian Abbott, the founder and CEO of Agentiq Capital, claimed in June that he suspected a Chinese employee he hired last year was acting as a Beijing operative. He alleged this individual intentionally manipulated code and website content to sabotage the company’s chances of securing venture capital. According to Abbott, the employee changed terminology from "ASI," or artificial superintelligence, to "fintech," a term that has lost its luster among investors. The employee has since been let go, and Abbott's company has filed a report with the FBI, although CNBC has not independently verified these claims. The FBI stated, "China's economic espionage campaign threatens the American economy, costing businesses hundreds of billions annually, and jeopardizing our national security." They emphasized their commitment to investigating any foreign attempts to steal U.S. technology.
While the Cyberspace Administration of China and the U.S. Department of State did not respond to inquiries, experts note the complexity in distinguishing between state-sponsored espionage and actions at the corporate or individual levels. Graham Webster, editor-in-chief of Stanford University's DigiChina Project, highlighted the influence of the current climate, where major U.S. firms prepare for significant public offerings. He remarked that "the narrative is overtaking reality in a lot of decisions," hinting at the government's attempts to curb China's advancements through technology export controls. He pointed out that it’s not surprising for the Chinese government to adopt contrary strategies.
Start-ups appear to be especially vulnerable to these threats.
