China-linked actors target more than technology as AI competition with U.S. intensifies

As the competition in artificial intelligence (AI) heats up, cyberattacks aimed at pilfering American technology are evolving. Recent reports indicate that actors linked to China are shifting their tactics from traditional tech-focused attacks to exploiting human vulnerabilities, marking a new chapter in the ongoing tech race. Matt Pearl, director of the strategic technologies program at the Center for Strategic and International Studies, emphasized that the People's Republic of China has increasingly targeted the tech sector. Rather than concentrating on specific trade secrets like hardware designs, these hackers are now interested in anything that could help them close the AI gap with the U.S., which currently stands at three to four months. Their focus ranges from uncovering a company’s product roadmap to pinpointing weaknesses within supply chains. The evidence of this growing trend is mounting. In June, cybersecurity firm CrowdStrike reported that Chinese entities were responsible for over half of state-sponsored cyber intrusions aimed at technology firms, particularly those involved in AI development, over the past year. Additionally, American startup Anthropic has accused Chinese corporations, including Alibaba, of attempting to unlawfully acquire its AI capabilities. Concerns about intellectual property theft aren't limited to established companies. U.S.-based AI content detection firm Copyleaks revealed that responses generated by Chinese startup DeepSeek’s R1 model showed striking similarities to those produced by OpenAI’s ChatGPT, leading to speculation about potential training on U.S.-developed models. Both DeepSeek and OpenAI did not respond to inquiries regarding these claims. Moreover, Brian Abbott, CEO of Agentiq Capital, alleged that a former employee hired from China manipulated code and website content to hinder the company's chances of securing venture capital. The employee, who was dismissed earlier this year, allegedly replaced terms related to artificial superintelligence with less favorable ones, such as “fintech.” The FBI has described China’s economic espionage as a persistent threat that costs the U.S. economy billions annually and jeopardizes national security. The bureau remains committed to investigating foreign attempts to steal American technology. Neither the Chinese Cyberspace Administration nor the U.S. Department of State offered comments on these allegations. Graham Webster, the editor-in-chief of Stanford University's DigiChina Project, noted the difficulty in distinguishing between state-sponsored espionage and individual or corporate initiatives. He pointed out that the narrative surrounding Chinese AI is heavily influenced by the imminent public offerings of major U.S. companies. As startups attempt to compete with tech giants, they often lack the necessary resources to defend themselves against such cyber threats, leading to what Cliff Steinhauer, director at the National Cybersecurity Alliance, referred to as “cyber poverty lines.” New employees are frequently targeted by attackers seeking access to valuable AI models, making human vulnerabilities a significant risk. In response to these challenges, companies like Anthropic are taking proactive steps. On June 11, they announced a program called Claude Corps, aimed at training 1,000 individuals in AI to support non-profits in the U.S. In contrast, China is ramping up support through initiatives that include subsidized computing resources and rent-free office space for startups. Isaac Stone Fish of Strategy Risks noted that while Beijing focuses primarily on larger corporations, startups are especially vulnerable due to their lack of cybersecurity expertise. He highlighted that in the past 18 months, China has intensified its efforts to ensure its companies lead the global AI race through various tactics, including cyberattacks and targeted government support for domestic competitors. For startups, the challenge remains to innovate rapidly while ensuring security. Abbott recounted that his former employee initially worked for free and was later compensated with a modest salary and stock options, underscoring the financial constraints many startups face in maintaining a secure environment.