A former Google employee has been accused of stealing highly confidential artificial intelligence technology from the company while working covertly for two Chinese companies, as per the Department of Justice’s announcement on Wednesday.
The indictment revealed that Linwei Ding took over 500 classified files concerning Google’s data centers used for training and hosting its extensive AI models.
Initially hired by Google in 2019 as a software developer for their supercomputing data centers, Ding allegedly misused his access to upload proprietary information to his personal account between May 21, 2022, and May 2, 2023.
Shortly after commencing the data theft in June 2022, Ding was reportedly recruited by a Chinese tech firm to act as its chief technology officer. From October 2022 to March 2023, while working remotely from China for the new company, Ding managed to deceive Google by persuading a colleague to swipe his access badge at the company building entrance.
Simultaneously, Ding established his own AI technology startup and assumed the role of CEO during this period.
The indictment further claims that Ding traveled to Beijing in late 2023 to secure financing for his company, pitching a plan to replicate and enhance a version of Google’s computational platform tailored for China’s specific requirements.
Google became aware of Ding’s actions when he transferred a set of files from the Google network to his personal account while in China. Subsequently, Ding swiftly resigned from Google by booking a one-way ticket to Beijing from San Francisco, following an inquiry by Google regarding the file transfer.
A Google spokesperson emphasized that this incident is isolated and not indicative of broader issues within the company. They assured that Google has robust measures in place to safeguard confidential information and trade secrets, promptly reporting the theft to law enforcement after an internal investigation.
Ding was apprehended in Newark, California, on Wednesday and is facing four charges of trade secret theft. If convicted, each count carries a maximum sentence of 10 years in prison and a $250,000 fine.
Authorities are concerned that foreign adversaries could exploit generative AI for covert influence campaigns to manipulate public opinion, especially during a presidential election year, potentially yielding significant advantages.
FBI Director Christopher Wray highlighted the increased threat posed by new technologies, enabling both sophisticated and unsophisticated foreign actors to engage in malign influence operations at a faster pace during the current election cycle.
