A San Francisco jury has delivered a historic verdict in what prosecutors call the first conviction for AI-related economic espionage, finding former Google software engineer Linwei Ding guilty on all 14 counts of stealing trade secrets related to the company's artificial intelligence infrastructure.
The conviction marks a watershed moment as nations race to dominate AI development. Ding, 38, systematically exfiltrated over 2,000 pages of confidential documentation detailing Google's supercomputing architecture—the very systems that power the tech giant's cutting-edge AI models.
Between May 2022 and April 2023, Ding exploited his privileged access to upload Google's crown jewels to his personal cloud account.
The stolen materials weren't abstract research papers—they contained granular specifications for Google's custom Tensor Processing Unit (TPU) chips, the orchestration software that binds thousands of processors into AI supercomputers, and proprietary SmartNIC designs enabling high-speed communication within data centres.
"Ding claimed he could build an AI supercomputer by copying and modifying Google's technology," according to court documents. He wasn't bluffing. While still employed at Google, Ding secretly positioned himself as CTO of a China-based tech startup and founded his own AI company, pitching investors on capabilities derived directly from stolen blueprints.
The timing raises red flags. Just two weeks before resigning in December 2023, Ding downloaded the stolen secrets to his personal computer. His talent plan application to the Shanghai authorities revealed his endgame: "help China to have computing power infrastructure capabilities that are on par with the international level."
"This conviction exposes a calculated breach of trust involving some of the most advanced AI technology in the world," said Assistant Attorney General John A. Eisenberg. The case underscores how insider threats—not just external hackers—pose existential risks to American tech leadership.
Ding faces up to 15 years per economic espionage count. His February status hearing will determine whether this landmark case sets a precedent strong enough to deter the next insider with access to tomorrow's technology.
For companies developing AI infrastructure, the message is unambiguous: your most sensitive assets may be walking out the door in someone's personal cloud account right now.