"A former Google software engineer has been convicted of stealing AI hardware secrets from the company for the benefit of two China-based firms, one of which he founded. The second startup intended to use these secrets to market its technology to PRC-controlled organizations. On Thursday, Linwei ("Leon") Ding, 38, was convicted on seven counts of economic espionage and seven counts of trade secret theft for stealing confidential information related to Google's Tensor Processing Units (TPUs), Graphics Processing Units (GPUs), and SmartNIC network interface cards."
"According to the second superseding indictment [PDF] filed in September 2025, Ding began working for Google on May 13, 2019. He was responsible for developing software that helps GPUs function more efficiently for machine learning applications at Google and Google Cloud. In that role, he had access to information Google deemed confidential. Google employs various physical and network security measures, including security guards, cameras at building entrances, badge-based building access restrictions, guest registration and accompaniment requirements, network data loss prevention monitoring, device identification and authentication, and network activity logging."
Linwei (Leon) Ding, 38, was convicted on seven counts each of economic espionage and trade secret theft for stealing confidential designs and information about Google's Tensor Processing Units (TPUs), Graphics Processing Units (GPUs), and SmartNIC network interface cards. Ding began working at Google on May 13, 2019, developing software to improve GPU efficiency for machine learning, which provided access to confidential data. He transferred stolen information to two China-based firms, including one he founded; the second startup planned to market the technology to PRC-controlled organizations. Google maintained layered physical and network security measures that later helped reconstruct Ding's activities but did not immediately detect the theft.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]