A recent congressional investigation has concluded that the Chinese artificial intelligence platform DeepSeek poses a serious and growing threat to U.S. national security.
According to a new report from the House Select Committee on the Strategic Competition Between the United States and the Chinese Communist Party, DeepSeek secretly harvests American user data, censors information according to CCP directives, and was likely built using stolen U.S. technology—all while relying on semiconductor chips that should never have reached China in the first place.
“DeepSeek represents a profound threat to our nation’s security,” the report warns. “Although it presents itself as just another AI chatbot… closer inspection reveals that the app siphons data back to the People’s Republic of China (PRC), creates security vulnerabilities for its users, and relies on a model that covertly censors and manipulates information pursuant to Chinese law.”
To address this threat, the report recommends an expansion of export controls, stronger enforcement against Chinese AI platforms, and the creation of a federal whistleblower program to report violations. It also calls for heightened coordination among national security agencies to prevent China from achieving a “strategic surprise” in the AI race.
How DeepSeek Works—and Why It’s Dangerous
The committee’s investigation found that DeepSeek transmits extensive data – including chat history, device details, and even user typing behavior – through back-end infrastructure connected to China Mobile, a state-owned telecom firm designated by the U.S. Department of Defense as a Chinese military asset. China Mobile has been banned from operating in the U.S. since 2019 due to fears that “unauthorized access to customer… data could create irreparable damage to U.S. national security.”
Cybersecurity analysts also discovered that DeepSeek sends user information with “no meaningful security measures,” raising serious concerns that the system was deliberately designed to make Americans’ data easily accessible to Chinese authorities.
Moreover, the AI model itself is manipulated to serve Beijing’s strategic interests. DeepSeek censored politically sensitive topics, including democracy, Taiwan, and human rights, in 85 percent of test cases. It doesn’t merely avoid controversial topics; it actively rewrites history and reinforces CCP talking points. “Beijing is not just censoring the internet at home. It is embedding its Great Firewall into platforms Americans use every day,” the report notes.
Built on Stolen Technology
DeepSeek’s capabilities didn’t come from scratch. Congressional investigators found that DeepSeek likely used “model distillation” – a technique that copies reasoning capabilities from other AI models – to replicate U.S. systems like OpenAI’s ChatGPT. OpenAI confirmed to Congress that DeepSeek employees circumvented protections, used fraudulent accounts, and extracted model outputs in violation of OpenAI’s terms of service.
“Through our review, we found that DeepSeek employees circumvented guardrails in OpenAI’s models… to accelerate the development of advanced model reasoning capabilities at a lower cost,” OpenAI told the committee. This kind of intellectual property theft poses serious challenges for U.S. companies trying to maintain a competitive edge in a high-stakes field.
Smuggled Chips Fuel the Engine
Even more disturbing is how DeepSeek built and trained its model. According to the report, DeepSeek uses tens of thousands of high-powered chips made by Nvidia, including A100s, H800s, and H100s – many of which are subject to strict U.S. export controls. These chips are crucial to building large-scale AI models, and selling them to China without a license is prohibited.
Yet DeepSeek appears to have acquired many of these chips through illegal channels. Investigators discovered a smuggling network operating out of Singapore, where three individuals – one a Chinese national – were charged for illegally exporting Nvidia chips to China. The network was busted shortly after members of Congress raised alarms about chip smuggling via Singapore.
This incident underscores a broader problem: U.S. companies and intermediaries are still supplying adversaries with the technological tools needed to match and surpass American capabilities.
A Pattern of Strategic Deception
DeepSeek’s founder, Liang Wenfeng, maintains effective control of the company through a complex corporate structure with ties to High-Flyer Quant, a firm that invested $420 million into DeepSeek’s development. The company operates within a state-subsidized Chinese tech corridor built to realize “Xi Jinping Thought” – the ideological core of the CCP.
Liang’s connections to military-linked researchers and the state-run Zhejiang Lab highlight how closely Chinese tech innovation is tied to national security goals. The app’s integration with entities like Tencent, Baidu, and ByteDance – each with their own histories of surveillance, censorship, or military affiliation – makes DeepSeek more than just a technological threat. It is, in effect, a tool of geopolitical warfare.
Lessons from the Cold War
Experts interviewed for this piece suggest the U.S. needs to reestablish the kind of strict export control framework that helped contain Soviet technological ambitions during the Cold War. Hans-Günter Förstner, a retired professor of international law who enforced Cold War-era trade controls for West Germany, recalled how the U.S.-led Coordination Committee for Multilateral Export Controls (COCOM) successfully denied Moscow access to critical technologies.
“The Americans… always emphasized that the priority was to deny access to all knowledge and products related to robotics and space technology,” Förstner said. “President Reagan was entirely correct.”
Dr. Xiàhóu Li Wei, a former senior CCP official who defected to the West, stressed that the current threat is even greater. “The CCP wanted the West to see China as a completely different entity from the Soviet Union,” he said. “It was a deception. The CCP kept winning until recently.”
Policy Recommendations: What Congress Must Do
To counter this growing threat, the report offers several urgent recommendations:
- Expand export controls to include new chip types like Nvidia’s H20 and improve enforcement mechanisms through whistleblower incentives and bilateral crackdowns on smuggling routes like Singapore.
- Require U.S. firms to track the end users of advanced chips and software.
- Mandate security and transparency standards for all AI systems trained on U.S.-origin technology.
- Prohibit the federal government from using Chinese AI platforms like DeepSeek.
The report concludes with a stark warning: “The potential for AI strategic surprise is most acute in the national security space. An AI weaponized and deployed by a U.S. adversary may prove to be a decisive advantage before a conflict starts”.
China’s AI ambitions are not just technological – they are ideological, strategic, and adversarial. The time to act is now.
Ben Solis is the pen name of an international affairs journalist, historian, and researcher.
Read full article here