Google AI Espionage: Former Engineer Guilty in High-Profile Data Theft for China Startup
- John Jordan

- 21 hours ago
- 3 min read
A former Google engineer has been found guilty by a federal jury of stealing confidential AI trade secrets to benefit technology startups linked to China. The case, considered one of the most significant examples of tech espionage involving US artificial intelligence, has sent ripples through Silicon Valley and highlights ongoing concerns over intellectual property theft.

Key Takeaways
Linwei Ding, a former Google engineer, convicted of economic espionage and theft of AI trade secrets
Ding stole thousands of confidential documents over nearly a year, including details on custom AI chips
The stolen information was intended to benefit his own China-based startup and Chinese government programs
Ding faces up to 10 years for each trade theft count and 15 years for each espionage count
The Theft: How It Happened
Linwei Ding, also known as Leon Ding, systematically copied over 2,000 highly sensitive files from Google. These files included intricate details of Google’s custom-built Tensor Processing Unit (TPU) chips, GPU systems, and the sophisticated software that powers Google’s AI supercomputers. Beginning in May 2022, Ding transferred proprietary information from Google’s internal network to his personal storage accounts. Authorities described his actions as a year-long operation, masking his tracks with tactics like converting files into PDFs and uploading them via various devices.
Business Ambitions in China
While still employed at Google, Ding actively sought opportunities in China’s rapidly growing AI sector. He engaged in discussions for a CTO role at a Chinese startup and eventually launched his own company, Shanghai Zhisuan Technologies Co., in 2023. Court documents reveal he pitched his ability to replicate Google’s AI supercomputing expertise to potential investors, positioning his firm at the forefront of China’s efforts to reach international standards in AI infrastructure.
Legal and National Security Implications
The jury in San Francisco delivered a swift guilty verdict on all counts of economic espionage and trade secrets theft after a twelve-day trial. Prosecutors stated Ding’s theft offered vast competitive advantages, potentially saving years of research and millions in development costs for his China-based initiatives. The case is emblematic of the US government’s heightened vigilance towards foreign attempts, particularly from China, to appropriate cutting-edge American technology.
U.S. prosecutors cited evidence showing Ding’s direct applications for Chinese government-sponsored talent programs, emphasizing his goal to boost China’s AI sector using the stolen know-how. While Ding’s defense argued that Google’s broad document access undermined the classification of the materials as true trade secrets, the jury disagreed.
What Happens Next?
Ding currently remains free as he awaits sentencing, scheduled for early February 2026. He could face a lengthy prison term—up to 10 years for each trade theft charge and up to 15 years for each economic espionage conviction.
The Broader Picture
This conviction shines a stark light on the increasing international competition surrounding AI technology. As the US and China race for dominance in the field, the case serves as a warning to both companies and individuals about the legal and ethical risks of intellectual property theft, as well as the strategic significance of AI in global affairs.
As cyber threats continue to evolve, your security strategy needs to evolve with them. BetterWorld Technology delivers adaptive cybersecurity solutions designed to keep your business secure while supporting innovation. Connect with us today to schedule a personalized consultation.
Sources
Ex-Google Engineer Convicted for Stealing 2,000 AI Trade Secrets for China Startup, The Hacker News.
Former Google Engineer Convicted of Stealing AI Secrets for China, Decrypt.
Ex-Google Worker Guilty of Stealing AI Tech for Chinese Firm (3), Bloomberg Law News.






