Summary:
Linwei Ding, a 38 year old Chinese national and former software engineer at Google, has been been indicted for illegally transferring over 500 confidential AI files from Google to his personal accounts. The stolen data, which includes critical hardware and software designs, was allegedly shared with Chinese companies in the AI sector. This includes advanced machine learning chips and Google’s proprietary Cluster Management System (CMS), which manages resources in Google’s supercomputing data centers. Ding is said to have begun transferring the data in May 2022 and continued until May 2023, all while covertly collaborating with two Chinese startups.

The indictment reveals that by no later than May 30, 2023, Ding had also secretly founded his own technology company in the AI and machine learning sector, where he served as CEO. His company promoted the development of a software platform aimed at accelerating machine learning tasks, including training large AI models. The indictment further alleges that Ding applied to a startup incubation program in China and traveled to Beijing in November 2023 to present his company at an investor conference. A document linked to his startup claimed that the company had experience with Google's computational power platform and intended to replicate and enhance it to suit China’s specific needs.

The indictment was unsealed following his arrest in Newark, California. Ding has been charged with four counts of theft of trade secrets. If convicted, Ding faces up to 10 years in prison and a $250,000 fine for each count.

Security Officer Comments:
According to the indictment, Ding used sophisticated methods to evade detection by Google’s data loss prevention systems. He allegedly copied source code to Apple Notes on his company-issued MacBook, converted the notes into PDFs, and uploaded them to a separate account. Furthermore, in December 2023, Ding allegedly let another Google employee use his access badge to create the illusion that he was working from a U.S. office while he was actually in China.

Suggested Corrections:
People are the first line of defense against Insider Threats. While there will be essential security awareness and training information that applies to all insiders (employees, partners, and contractors), you should strive to tailor it to the tasks of their specific roles and accesses. The goal should be to take your users beyond mere awareness of security policies and issues and truly educate them. They should be instructed on why and how to assess various situations' risk and security implications. You should verify that they know how to apply security best practices as they perform their daily job duties.

Insiders should have access to only those information assets for which they 1) have a need-to-know based on the role and duty and 2) that fall within the parameters of their risk profile.

The employee monitoring market consists of technologies that collect data about the location, movement, communications, and actions of employees. Because of their narrow focus, these tools are often integrated with other tech stack tools to support broader purposes. The most prevalent use cases for EM products are optimizing employee, team, and process productivity and efficiency by tracking physical and electronic activities and reducing bandwidth costs emanating from the inappropriate use of devices and networks.

Organizations should strive to examine behaviors, actions, and insider threat indicators to mitigate threats following established policies, existing business objectives, risk tolerance, and legal parameters. Typical data loss, fraud, and intellectual property are the framework for investigative best practices; however, they can often miss critical data sources. Integrating evidence from HR, Legal, Cyber, and Security is essential consideration with investigative methods.

The IT-ISAC has an excellent resource on Insider Threats (Insider Threat Starter Guide): https://130760d6-684a-52ca-5172-0ea...d/b8fa6c_74013b122d234c16b07eeb52af6fdd42.pdf

Link(s):
https://www.justice.gov/archives/op...d-theft-artificial-intelligence-related-trade