Chinese Cyber Espionage Targets U.S. AI Startups Through Insiders

Espionage tactics shift from networks to people

Cyber espionage aimed at American artificial intelligence technology is taking a new form. Rather than relying solely on technical network breaches, attackers—particularly those linked to China—are increasingly exploiting human vulnerabilities within organizations, according to security experts and company executives.

Matt Pearl, director of the strategic technologies program at the Center for Strategic and International Studies, said Chinese actors have broadened their focus beyond specific trade secrets like hardware designs. They now target anything that could close what he estimates is a three- to four-month AI development gap with the United States, including product roadmaps and supply chain weaknesses.

CrowdStrike reported in June that Chinese entities were responsible for more than half of state-sponsored intrusions targeting technology companies and their AI assets in the twelve months ending March 31, as first detailed by CNBC.

Startups face disproportionate risk

Small AI companies are particularly exposed. Cliff Steinhauer, director of information security at the National Cybersecurity Alliance, described "cyber poverty lines" where startups lack the defensive resources available to large corporations. Social engineering tactics, amplified by AI-generated content, exploit this gap.

Alon Yamin, CEO of AI detection startup Copyleaks, said new employees at his company are immediately targeted by cyberattacks seeking access to AI models. He expects such incidents to increase.

Brian Abbott, founder of startup Agentiq Capital, told CNBC he believes an employee hired from China in the previous year was a state agent who deliberately sabotaged code and website content to obstruct venture capital funding. Abbott said the individual replaced references to "artificial superintelligence" with "fintech," a term that has fallen out of favor with investors. The employee was terminated and the company filed an FBI complaint, though CNBC could not independently verify the allegation.

Attribution challenges and broader context

Graham Webster, editor-in-chief of Stanford's DigiChina Project, noted that distinguishing state-sponsored espionage from individual or corporate theft remains difficult. He also suggested that narratives around Chinese AI threats may be influenced by U.S. companies preparing for major public offerings.

Isaac Stone Fish, CEO of consultancy Strategy Risks, said Beijing's efforts have intensified over the past eighteen months. He described a multipronged approach including supply chain restrictions, employee harassment, hacking, and subsidies for Chinese competitors.

The FBI confirmed to CNBC that China's economic espionage campaign costs the American economy hundreds of billions of dollars annually and poses national security risks.

Why it matters

The shift from purely technical attacks to insider threats fundamentally changes the security calculus for AI companies. Startups racing to compete or position themselves for acquisition often prioritize speed over security infrastructure, creating systematic vulnerabilities that well-resourced state actors can exploit. As AI becomes central to economic and military competitiveness, the human element—from hiring practices to employee vetting—may prove as critical as network defenses.

These details were first reported by CNBC.