The Unveiling of a Cybersecurity Breach
During the excitement surrounding the launch of China’s first AI model by the company “DeepSeek,” the U.S. Department of Justice accused former Google engineer Lin Wei Ding of stealing AI secrets from the company and handing them over to Chinese firms. This marks the beginning of a new economic war in the AI sector between China and the U.S.
Background Investigation and the Biden Administration’s Response
The accusation against Ding follows an extensive investigation by several federal agencies, driven by a joint task force established by the Biden administration in 2023 to combat disruptive technologies aimed at stealing and sabotaging American corporate secrets. Although the task force was established during Biden’s tenure, Trump maintained and relied on it during his election campaign, promising to curb China’s access to American technologies and prevent espionage.
The Theft of 1000 Confidential Documents from Google
Lin Wei Ding, also known as Leon Ding, joined Google in May 2019 as a member of their software development team. His espionage activities began three years later, specifically from May 2022 to May 2023, when his theft was discovered. During this period, Ding managed to steal and upload over 1000 confidential documents from Google to his personal Google Drive account, including several training materials used by the company for new employees.
Focus on Google’s AI Chips
The stolen documents primarily revolved around Google’s AI chips, which the company aimed to use to compete with Amazon and Microsoft’s chips. This was part of their efforts to move away from Nvidia chips and reduce reliance on them. These documents included detailed designs of the new chips and their operational mechanisms, particularly the “Tensor Processing Units” (TPUs) that the company heavily depends on for their Pixel smartphones. These TPUs have also been integrated into Google’s data centers to power their supercomputer.
In addition to the design documents, the stolen materials contained detailed information about the operational mechanisms and the necessary API interfaces to train the company’s supercomputer models and various AI models, including Gemini and their image generation model, according to the official indictment statement by the U.S. Department of Justice.
The Role of DeepSeek and Its Alleged Involvement
Currently, DeepSec faces allegations from the U.S. government and OpenAI for using ChatGPT to train its AI model “R1.” Despite intensive investigations by OpenAI, no official accusation has been made against the Chinese competitor. Similarly, in Ding’s case, it appears that DeepSec has no direct involvement.
The Department of Justice report indicates that Ding obtained a prestigious technical position at a Chinese AI startup during the same period he was stealing documents from Google. Furthermore, while still employed at Google, Ding traveled to China to secure necessary funding for the company.
Establishing a New AI Company in China
In May 2023, Ding established a second tech company in China focusing on AI and machine learning. He served as its CEO while still working at Google and continued uploading confidential documents to his account until December 2023. His home was subsequently raided by the FBI for investigations, and he was arrested in March 2024.
Charges Against Lin Wei Ding
According to the Justice Department’s statement, Ding faces 14 charges of economic espionage and theft of trade secrets—seven charges for each company he collaborated with while employed at Google. The federal grand jury in San Francisco has indicated that each charge of economic espionage carries a maximum penalty of 15 years in prison and a fine of up to $5 million. Each charge of trade secret theft carries a maximum penalty of 10 years and a fine of $250,000.
Despite denying all charges, Ding continues to fight the case, having been convicted on four counts in March. He is currently out on bail awaiting the final verdict on the remaining charges.
Attempted Exploitation of Beijing’s Talent Support Program
Ding also attempted to join Beijing’s Talent Support Program, which offers financial rewards for bringing technical expertise from abroad. In his application, he claimed he could help China achieve global competitiveness in AI technologies and offered his services to the Chinese government and universities.
Discovery of the Theft by Google
Google discovered Ding’s theft during an AI conference in Beijing, where he claimed to be the CEO of a startup while still employed at Google. This prompted the company to initiate an internal investigation, leading to the collection of incriminating evidence.
Google’s Cooperation with Federal Authorities
The Attorney General has not filed any charges against Google, as the company has closely cooperated with federal agencies in convicting Ding. Additionally, Google has reinforced its security systems to prevent similar leaks in the future.
Potential Risks of the Stolen Documents
Beyond the security risk posed to Google’s supercomputers and data centers—where the stolen documents could potentially enable the hacking of any data center relying on the mentioned technologies—they also reveal Google’s plans to competitors, even if the documents do not reach Chinese firms.
Moreover, these documents could help any company establish its own AI systems and build advanced chips without purchasing them directly from specialized companies. The U.S. government is striving to prevent China from achieving this by all means possible in recent times.
