Nvidia scandal: The AI arms race puts human spies back into the game

Nvidia's decision to launch an independent investigation into CEO Jensen Huang's reported relationship with a woman, Alice Huang, a former executive of Megaspeed, suspected by US intelligence of spying for Beijing has brought an uncomfortable question to the fore. Is the AI industry becoming a new battleground for human espionage? Ironically, when AI tools are supposed to replace the old-style spycraft, human spies are back in the game in the Silicon Valley.

The Nvidia controversy follows the arrest of a former OpenAI engineer accused of planting a concealed listening device in an executive conference room at the company's San Francisco headquarters. It also comes after a former Google engineer was convicted of stealing confidential AI technology.

ALSO READ | Alice in Nvidia CEO's secret world: Jensen Huang's personal affair matters far beyond than what meets the eye


These incidents suggest that the AI revolution has not made traditional espionage obsolete. If anything, it may have given human intelligence a new lease of life.

The human spy in the age of AI

The OpenAI case offers a striking illustration of how human intelligence is irreplaceable even in the AI industry. Abdulrahman Alfozan, a Saudi national who had worked as an engineer at the company since July 2025, was arrested last month and charged with attempting to intercept oral communications. According to the FBI complaint, a concealed recording device was discovered inside a USB port wrapped in toilet paper at the back of a television in an executive conference room. Investigators also found another similar device in his bag.

ALSO READ | I was unfaithful to my wife: Nvidia CEO Jensen Huang admits secret affair with Chinese spy Alice Huang

The room was regularly used for discussions involving confidential research, product plans, commercial opportunities and government engagements. Prosecutors alleged that Alfozan had used an AI chatbot to research recording devices and obtain information about OpenAI's meeting rooms and executive schedules. OpenAI said it had found no evidence that he accessed or extracted model weights, source code or customer data. The allegations have not yet been established in court.

The case is unusual, but the underlying objective is the old-style obtaining of information by gaining access to people and conversations rather than breaking into computer systems.

The Nvidia controversy brings another dimension to this. US intelligence agencies believed Alice Huang, a former executive at Singapore-based data centre company Megaspeed, was working for Beijing. She was reportedly suspected of cultivating a close relationship with Jensen Huang to gather information about Nvidia's strategy and its efforts to persuade Washington to relax restrictions on advanced chip exports to China. Huang has admitted to having a relationship with her.

The report has raised questions about the vulnerability of corporate executives whose relationships extend across geopolitical fault lines. However, the allegations remain unverified publicly. Singapore has also clarified that its government neither investigated Alice Huang nor raised concerns about her alleged intelligence connections, while confirming that investigations into Megaspeed are ongoing.

AI has changed the economics of espionage, not its human element

The case of former Google engineer Linwei Ding offers another insight into the problem. In January 2026, a federal jury found him guilty of stealing thousands of pages of confidential Google documents relating to AI technology. The material included information about Google's custom tensor processing units and computing infrastructure. Prosecutors said he had transferred the material while pursuing business interests in China.

However, a federal judge later ruled that prosecutors had failed to establish the necessary intent to benefit the Chinese government. The economic-espionage convictions were overturned, although the trade-secret theft convictions remained intact.

The case nevertheless highlights the importance of insiders. Unlike an external hacker, an employee with legitimate access can potentially obtain sensitive information without having to defeat the organisation's entire security infrastructure.

This is where AI presents an interesting paradox. AI agents can render human spies obsolete by executing flawless, high-speed cyber espionage that eliminates human vulnerability. Operating at digital scale, these autonomous systems can exploit zero-day vulnerabilities, bypass biometric security via hyper-realistic deepfakes, and silently exfiltrate classified data without the physical risks, emotional liabilities, or threat of defection inherent to human operatives.

This has dramatically reduced the cost of obtaining information. Algorithms can sift through enormous volumes of communications and public records to identify patterns that might otherwise escape humans.

But AI agents cannot independently retrieve a confidential conversation that was never recorded or discover an executive's private intentions simply by analysing stored information.

In fact, as AI technology becomes more strategically valuable, the importance of the people controlling it could increase. A handful of executives and senior engineers possess knowledge about future products, research priorities, government negotiations and commercial partnerships that competitors cannot easily obtain elsewhere. Human intelligence, therefore, is not necessarily being displaced by technological intelligence. The two can reinforce each other.

An AI espionage arms race

The emerging pattern has implications beyond corporate security. The US and China are competing for technological advantages that could shape their economic strength and military capabilities for decades. Advanced semiconductors, computing infrastructure and frontier AI models have consequently become matters of national security.

America's restrictions on Nvidia's advanced chips reflect its concern that access to powerful computing technology could strengthen China's military and intelligence capabilities. China, meanwhile, is investing heavily in domestic alternatives and pursuing greater semiconductor self-reliance.

The contest has echoes of the Cold War nuclear rivalry. Then, the superpowers competed not just to build more powerful weapons but to acquire information about each other's technological capabilities. Scientific secrets, research facilities and the people working inside them became intelligence priorities.

The AI competition could produce a similar dynamic, although the technologies and strategic calculations are different. Unlike nuclear weapons, AI is being developed largely by private companies whose commercial interests do not always align with government security priorities. This makes the intelligence landscape more dispersed and difficult to police.

There is also a new complication. AI itself can assist espionage by analysing stolen material, identifying potential sources and generating convincing digital identities. But the final breakthrough may still depend on persuading a human being to share information or gaining access to a room where important decisions are being discussed.

The Nvidia scandal comes right before Trump-Xi meet

The timing of the Nvidia controversy is particularly sensitive. It has surfaced just before US President Donald Trump is expected to meet Chinese President Xi Jinping at the upcoming APEC summit in China. The two governments are trying to preserve a fragile trade understanding. China has extended its suspension of certain trade measures for another two months, keeping the negotiations alive.

Nvidia sits close to the centre of this diplomatic contest. Its chips are essential to the global AI industry, while Jensen Huang has been lobbying Washington for greater access to the Chinese market. Megaspeed, the company previously associated with Alice Huang, reportedly purchased $4.8 billion worth of Nvidia chips over two years. It has also faced scrutiny over allegations of helping restricted chips reach China.

The intelligence allegations could complicate an already difficult negotiation. The US may face renewed pressure to scrutinise chip exports and the relationships between American technology executives and foreign customers. China, which has dismissed the allegations, could view the controversy as another attempt to undermine its technological ambitions ahead of the summit.

There is no evidence that the reporting was deliberately timed to influence the negotiations. Nor is there any public proof yet that Nvidia's secrets or US national security were compromised. The company's board has said it has no reason to believe either happened, while announcing an independent inquiry.

But the larger issue will outlast this particular controversy. As AI becomes a source of economic and strategic power, the people who develop and control it will attract growing attention from intelligence agencies and commercial rivals.

AI agents may be transforming how computers are hacked to access secret information, but they have not changed the fundamental value of knowing what another human being knows. In the emerging AI arms race, the human spy may remain a potent weapon.