Former Google Engineer Stole AI Secrets and Built a Startup in China
An engineer who worked at Google was convicted of stealing artificial intelligence secrets and using them to build a startup in China. The case came to light during a hearing before the Senate Judiciary Committee, where a former CIA officer gave detailed testimony about how this story goes far beyond a simple corporate crime.
What was at stake was nothing less than the national security of the United States and the future of the global tech race. And the timing could not have been more sensitive.
Artificial intelligence has become the number one priority on the American government’s agenda. President Donald Trump has made AI a centerpiece of his policy, pushing for a single federal regulatory framework instead of a patchwork of different state laws. His administration has also been pressing to speed up the construction of data centers and strengthen US competitiveness against China.
Cases like this show that the tech rivalry with China is far from a fair fight between ordinary competitors. According to experts who testified before the Senate, American companies are not competing against market rivals — they are competing against the largest intelligence apparatus in the world 🌐.
In this article, you will learn who the engineer involved is, what exactly was stolen, how that material became fuel for a company in China, and why this episode has become a landmark in discussions about economic espionage and intellectual property in tech.
Who Is the Engineer at the Center of the Case
The engineer’s name is Linwei Ding, also known as Leon Ding. He is a Chinese national who worked at Google as a software engineer and, during that time, had direct access to some of the company’s most sensitive systems. That included proprietary architectures related to chips and artificial intelligence infrastructure. In January, federal prosecutors confirmed that Ding was convicted on multiple counts of economic espionage and trade secret theft, after stealing thousands of pages of confidential AI-related information for the benefit of China. The case was tried in a federal court in California and stands as one of the first major convictions in the United States tied to espionage in artificial intelligence.
According to evidence presented at trial, Ding downloaded sensitive data about Google’s AI infrastructure, including chip designs and software used to train advanced models. He then sent this material to a personal account while secretly collaborating with China-based tech companies. To cover his tracks, he used a method that, despite being simple, proved effective for quite a while: he transferred files to personal devices and then moved them to his own cloud storage. The company discovered the activity after noticing unusual patterns in internal access logs, which led to an investigation referred to the FBI.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The detail that made the case even more serious was the fact that, while still employed at Google, Ding had already founded companies in China and launched his own startup. Prosecutors said he sought to use the stolen technology to build AI systems in China and attract investors, highlighting the case as part of a broader effort by Beijing to acquire advanced technology from the United States. In other words, he was building direct competition using material being extracted from his employer, and he was doing it in a structured way, with financial and institutional support coming from China.
What Was Stolen and Why It Matters So Much
The stolen files were not generic documents about AI. They included technical details about Google’s supercomputing infrastructure, specifically about the design of TPU chips (Tensor Processing Units). These processors were developed by the company to accelerate the training of artificial intelligence models at scale. TPUs represent years of research and billions of dollars in development and are considered one of Google’s key competitive advantages in the global AI market. Having access to this technology means drastically shortening the path to replicating capabilities that took a decade to build.
Beyond the hardware data, the files also contained information about the software that manages these systems, including internal frameworks and libraries used to train large-scale language models — the now-famous large language models. This information is extremely valuable because it reveals not only what Google built, but how the company thinks about systems architecture, which problems were prioritized, and which solutions were considered most efficient. It is basically the treasure map of modern AI research, and that map ended up in the hands of people who wanted to use it to compete directly with the Americans.
From an intellectual property standpoint, the impact is enormous. When a company invests in research and development for years, it builds an advantage that goes far beyond the final product that hits the market. It is embedded in the processes, the architecture decisions, the mistakes that were avoided along the way. Stealing these secrets is like skipping stages of a race without actually running. And when that material ends up with companies operating in a different ecosystem, with different rules and without the constraints that the Western market imposes, the competitive imbalance becomes even harder to reverse.
The Testimony That Shook the US Senate
Tom Lyons, a former CIA officer with more than 20 years of experience in the US government and the private sector on issues of Chinese economic espionage, delivered the most hard-hitting testimony during the hearing. He told senators that the playing field is completely tilted against American companies.
American companies are not competing against Chinese rivals in any normal sense, Lyons stated. They are competing against the largest intelligence apparatus in the world, whose mission includes putting American companies out of business.
Lyons made a point of being crystal clear that this kind of rivalry has nothing to do with traditional competition between large corporations. This is not GM versus Ford, he told lawmakers in his opening remarks. This is an American startup against the resources of the Chinese military, the People’s Liberation Army.
The strongest point in his testimony was his criticism of how the US government has been handling the issue. Lyons warned that the current approach leaves companies essentially on their own to face state-sponsored threats, treating what should be a national security matter as a corporate compliance problem.
If a foreign army were conducting operations on American soil, we would not ask our companies to fund their own defense, Lyons said, summing up what he considers an absurdity in how the United States deals with this threat.
Economic Espionage and National Security Take Center Stage in the Senate
The hearing before the Senate Judiciary Committee was a turning point in discussions about how the United States has been handling the threat of economic espionage in strategic areas of technology. Lyons’ testimony highlighted that Ding’s case is not an isolated incident but part of a systematic pattern of technological intelligence gathering. This pattern involves state agents, private companies, and individuals recruited into strategic positions within major Western corporations. The message was clear: the problem is structural and it is being underestimated.
What makes this discussion even more complex is the intersection between national security and technological innovation. The United States is the country that attracts the most global tech talent, and a large part of the progress made by major AI companies depends precisely on this diversity of people and perspectives. Creating overly rigid barriers could hurt the very ecosystem it is meant to protect. On the other hand, ignoring the risks of infiltration in sensitive areas is, as the case demonstrated, a decision that is far too costly. The challenge lies in finding a balance that protects critical assets without stifling the ability to innovate.
US authorities have argued for years that the theft of intellectual property by China has cost the American economy billions of dollars in revenue and thousands of jobs, representing a significant risk to national security. China, for its part, has repeatedly denied involvement in this type of activity.
The Role of Artificial Intelligence in the Geopolitical Rivalry
In the context of the tech race between the US and China, artificial intelligence plays a central role that goes beyond the marketplace. It is directly linked to the development of military capabilities, surveillance, large-scale data analysis, and real-time strategic decision-making. When secrets in this field cross borders illegally, the implications go far beyond financial losses for private companies.
That is why the US government has been toughening its stance in investigations into this type of activity, with the FBI and the Department of Justice prioritizing cases involving technology transfer to countries considered strategic adversaries 🔐. Linwei Ding’s conviction is a clear signal that tolerance for this kind of action is shrinking and that American courts are willing to hand down severe sentences in cases of tech espionage.
This landscape also directly impacts the startup and AI investment ecosystem. Venture capital firms funding artificial intelligence projects now have even more reason to demand rigorous due diligence on the backgrounds of collaborators and the protection of data circulating within organizations. Trust is an asset as valuable as the code itself, and episodes like this shake that trust in a lasting way.
What This Episode Reveals About the Global Tech Rivalry
Ding’s case has become a symbol of something that experts in tech geopolitics have been pointing out for years: the competition between major powers for dominance in artificial intelligence does not happen only in research labs or on stock exchanges. It also happens in corporate hallways, in access to internal systems, and through recruitment networks that operate quietly but in a highly organized fashion. The sophistication with which the material was extracted over months, without raising immediate red flags, shows that this is not the kind of threat you solve with a more powerful firewall.
For tech companies, the episode reinforces the need to rethink data access policies for sensitive information, especially in projects involving high-value strategic intellectual property. This includes not only monitoring for unusual behavior in systems but also a more careful review of onboarding processes for new employees, the permissions granted to team members at different career stages, and offboarding practices when someone leaves the company. In Google’s case, detection was possible — but not before a significant amount of material had already been transferred.
The Future of Protecting Secrets in Artificial Intelligence
From a regulatory perspective, this episode fuels debates about how governments should act to protect strategic sectors without setting precedents that could be misused. The line between legitimate oversight and overreach is thin, and any public policy in this area needs to be built carefully. Trump’s proposal for a single federal regulatory framework for AI could be a first step in that direction, but there are still many details to be worked out regarding how this framework will specifically address the protection of trade secrets and economic espionage.
What is clear after this case is that economic espionage in the tech field is a present reality, not a future threat. The response to it requires both technical intelligence and legal clarity to be effective. And above all, it requires a shift in mindset: protecting innovation in artificial intelligence is not just a market issue — it is a matter of technological sovereignty 🧠.
For anyone following the tech industry, this case serves as a powerful reminder that the most impressive advances in AI come with equally impressive risks. The next major battle in artificial intelligence may not take place in a performance benchmark or a product launch. It may happen in a courtroom, at a Senate hearing, or unfortunately behind the scenes at a company where someone decided that privileged access to confidential information was an opportunity too good to pass up.
