U.S. House Targets China’s DeepSeek Over Data Use

DeepSeek AI Under Fire: U.S. Lawmakers Call It a 'Profound Threat'

The AI battle between the U.S. and China just took a heated turn. A recent bipartisan report from the U.S. House Select Committee on China has labeled China’s DeepSeek AI as a serious national security risk. The reason? The report claims the company is funneling American user data back to China and that's not all.

As Washington ramps up pressure on Beijing’s tech ambitions, DeepSeek has become the latest flashpoint in an escalating AI cold war between the world’s two largest economies.

What’s DeepSeek and Why Is It in Trouble?

DeepSeek is a Chinese artificial intelligence company that’s been making waves for its affordable open-source models, especially R1, which can be hosted on U.S. servers. The model gained traction among developers and enterprises for its impressive reasoning abilities, at a fraction of the cost of U.S. counterparts.

But according to lawmakers and OpenAI, DeepSeek didn’t build its models from scratch. Instead, they allege the company used "unlawful" training methods, including a process called distillation, to essentially mimic OpenAI’s ChatGPT responses and internal logic.

Distillation isn't new in the world of AI. It's a known technique where a smaller model learns from the outputs of a larger one. But OpenAI says DeepSeek took things further, bypassing safety guardrails and even using OpenAI’s tools to grade and filter training data.

U.S. House Committee Sounds the Alarm

In a strongly worded report, the House committee stated:

“DeepSeek poses a profound threat to national security by collecting user data on Americans and sending it back to China.”

This claim is particularly serious, as it touches on surveillance, state-sponsored data collection, and concerns that the AI models could be subtly aligned with the interests of the Chinese Communist Party.

The report also pointed to the dual leadership of Liang Wenfeng, who runs both DeepSeek and a hedge fund called High-Flyer Quant, which is allegedly connected to state-linked research labs in China.

Fallout: Export Restrictions and Procurement Bans

Following the report, lawmakers are urging the U.S. government to:

• Ban the export of AI models to China, including open-source models like R1.
• Prohibit federal agencies from purchasing or using AI tools from Chinese companies.
• Investigate American chipmakers, like Nvidia, for potentially overstepping export rules.

While many of these actions might sound symbolic (since U.S. agencies don’t exactly line up to buy Chinese AI), the broader implications for AI research, global tech trade, and chip exports are significant.

Nvidia Hit by Trump’s Policy Shift

The timing couldn’t be worse for Nvidia. Just one day before the House report dropped, the Trump administration imposed fresh restrictions on chip exports to China, including the popular but previously exempt Nvidia H20 processors.

These chips were specially designed to skirt earlier export bans, offering limited AI performance but remaining legal for sale in China. Nvidia made billions from them.

Now, licenses will be required even for these lower-end processors, a move some experts criticize as self-defeating.

“If China can't buy American chips, it will just make its own and faster,” one analyst noted. “That’s exactly what happened with Huawei.”

Hypocrisy in the AI World?

The irony of the situation hasn’t gone unnoticed.

While OpenAI accuses DeepSeek of training on its content, many argue OpenAI itself built ChatGPT by scraping the open web, including copyrighted and proprietary material, without asking first. Critics call it a classic case of “pot calling the kettle black.”

Still, OpenAI’s concerns about intellectual property theft and AI ethics are resonating with U.S. policymakers, who fear that China could weaponize AI for propaganda or surveillance.

Can the U.S. Really Stop China’s AI Rise?

Not likely.

Despite growing bans and export restrictions, China has repeatedly proven its ability to innovate under pressure. When Huawei was cut off from Western semiconductors, it accelerated its push to build domestic chips. The same could happen with AI hardware and software.

Even if R1 was trained using OpenAI's methods, it’s now open-source and hosted globally by companies like Microsoft, Meta, and Perplexity. That makes it difficult to contain or ban outright.

Meanwhile, anyone accessing chat.deepseek.com is using servers in China but that alone doesn’t make the tool dangerous. It’s the data handling and training process that matters most, and that’s where U.S. lawmakers want answers.

TikTok, Tariffs, and Tensions

The report on DeepSeek also follows renewed efforts to force TikTok's U.S. operations to break away from its Chinese parent, ByteDance. However, Beijing has shown no interest in negotiations until tariffs imposed by the Trump administration are lifted.

It’s a classic political standoff: tech vs tariffs, national security vs market access.

In-Depth: What’s at Stake?

The U.S. fears go beyond just data leaks. Here's what’s really at risk:

• Model Manipulation: AI tools trained in China could be subtly skewed to reflect pro-CCP narratives or limit certain topics, posing ideological risks.
• Military Use: AI models could be adapted for defense or surveillance, accelerating China’s goals in cyber warfare and foreign intelligence.
• Economic Disruption: As U.S. firms like Nvidia, Microsoft, and OpenAI are limited in what they can sell, China could gain self-sufficiency faster and eventually become a global AI competitor.

Final Thoughts

The U.S.-China AI rivalry is just heating up and DeepSeek has landed squarely in the middle of it. While the House panel's report underscores real national security concerns, some of its recommendations could have unintended consequences: fueling China's tech independence and cutting off billions in U.S. chip sales.

Whether DeepSeek becomes a cautionary tale or a catalyst for change, one thing is clear: the AI race is now deeply entangled with politics, power, and profit.

Got a hot tech tip or insider scoop? Share it with our editorial team at [email protected] – we’d love to hear from you.