Ex-Google Engineer Arrested, Charged With Stealing AI Trade Secrets for China
The Chinese language nationwide boasted concerning the potential of his Chinese language startup to adapt Google’s computational energy platform for China’s nationwide circumstances.
A former Google software program engineer has been indicted on fees of stealing synthetic intelligence-related commerce secrets and techniques whereas working for Chinese language opponents in search of to realize an edge within the AI race.
Linwei Ding, a Chinese language nationwide who additionally goes by Leon, was charged by a federal grand jury in San Francisco with 4 counts of commerce secret theft, every punishable by as much as 10 years in jail. He was arrested on March 6 morning in California’s Newark metropolis, the place he lives.
The 38-year-old allegedly stole over 500 information containing confidential data between Could 2022 and Could 2023, together with detailed details about the {hardware} infrastructure and software program platforms permitting Google’s supercomputing information facilities to coach massive AI fashions via machine studying, in response to the indictment.
Inside weeks after Mr. Ding started the theft exercise, the indictment mentioned, an early-stage Chinese language firm with a give attention to AI supplied to make him its chief expertise officer. The place got here with a month-to-month wage of round $14,800 with an annual bonus and firm inventory.
That late October, Mr. Ding traveled to China and stayed there till the next March, collaborating in investor conferences to boost capital for the agency, Beijing Rongshu Lianzhi Expertise.
In Could 2023, he based an AI start-up in Shanghai.
“We’ve expertise with Google’s ten-thousand-card computational energy platform; we simply want to copy and improve it—after which additional develop a computational energy platform suited to China’s nationwide circumstances,” he said in a doc selling his firm on the China-based social media platform WeChat.
He additionally tasked one other Google worker to scan his entry badge on three separate days in December 2023 to create the impression that he was working from the U.S. Google workplace when, in truth, he was in China, the Google investigators discovered after analyzing surveillance footage.
Mr. Ding initially managed to evade Google’s detection by copying the Google information into the Apple Notes software on his Google-issued MacBook then changing it into PDF to add into his private Google Cloud account.
However in December 2023, when he uploaded further information from the Google community to a different private account whereas in China, Google turned suspicious.
He then advised a Google investigator that he had meant to make use of the data as proof of the work that he had carried out at Google, in response to the indictment.
Mr. Ding didn’t disclose both of his affiliations with China-based companies with Google, prosecutors mentioned.
Lower than every week later, he booked a one-way ticket to Beijing slated to depart on Jan. 7. He then resigned from Google on Dec. 26.
Google retrieved Mr. Ding’s Google laptop computer and cell gadget from his residence the day earlier than his deliberate departure on Jan. 4.
Stealing US Innovation
Lawyer Basic Merrick Garland disclosed the main points of the case at an American Bar Affiliation Convention in San Francisco on Wednesday afternoon.
“We’ll fiercely shield delicate applied sciences developed in America from falling into the fingers of those that mustn’t have them,” he mentioned.
FBI director Christopher Wray mentioned the costs have been “the newest illustration of the lengths associates of firms based mostly within the Folks’s Republic of China are prepared to go to steal American innovation.”
“The theft of progressive expertise and commerce secrets and techniques from American firms can value jobs and have devastating financial and nationwide safety penalties,” he mentioned in an announcement.
Deputy Lawyer Basic Lisa Monaco in a speech final month mentioned that AI is a prime precedence for the duty power, describing it because the “final disruptive expertise.”
Mr. Wray, in a late February nationwide safety convention, additionally warned concerning the hazard of generative AI in “making it simpler for each extra and less-sophisticated overseas adversaries to interact in malign affect” and interference with the U.S. political course of.