Chinese State-Linked Accounts Promote DeepSeek’s Launch on Social Media: Report

The icon for the smartphone app DeepSeek on a smartphone screen in Beijing on Jan. 28, 2025. Andy Wong/AP Photo

A recent report by Graphika showed that social media accounts linked to Chinese state agencies were pushing narratives favoring the Chinese artificial intelligence (AI) DeepSeek just days before the plunge in U.S. chip stocks occurred.

The analysis, obtained by Reuters, found that Chinese diplomats, embassies, and state media amplified news coverage of DeepSeek’s newly launched AI model, with the messaging that the app could threaten U.S. dominance in the AI industry.

These posts were shared across multiple Chinese social media platforms, including U.S.-based ones such as X and Facebook, just days before the app’s launch sent shockwaves through the U.S. tech market, according to the report.

Graphika said it observed a slight increase in discourse about DeepSeek’s progress and OpenAI’s ChatGPT on X following the launch of the DeepSeek AI model on Jan. 20. The discussion comparing the two continued to grow just days later, the report stated.

Jack Stubbs, chief intelligence officer at Graphika, said this demonstrated China’s ability to mobilize actors “that seed and amplify online narratives casting Beijing as surpassing the U.S. in critical areas of geopolitical competition, including the race to develop and deploy the most advanced AI technologies.”

DeepSeek last week launched its free, open-source model DeepSeek-R1, which it claimed is on par with OpenAI’s reasoning model called o1. The app topped Apple’s App Store in the United States and caused U.S. chip stocks to plummet, with Nvidia’s shares dropping by about 17 percent on Jan. 27.

DeepSeek announced that its AI models were developed at a fraction of the cost of its rivals. One of DeepSeek’s research papers stated that it used about 2,000 of Nvidia’s H800 chips to train its AI models, much less than the 16,000 chips typically used by other companies.

But some developers are skeptical about the company’s claim. Scale AI founder and CEO Alexandr Wang told CNBC on Jan. 23 that it is possible that DeepSeek has about 50,000 Nvidia H100 chips but could not disclose this information due to U.S. export controls—which ban the sale of advanced AI chips to China.

OpenAI told The Epoch Times on Jan. 29 that DeepSeek might have “inappropriately distilled” its models and is reviewing the evidence.

“We know that groups in the PRC are actively working to use methods, including what’s known as distillation, to try to replicate advanced U.S. AI models,” an OpenAI spokesperson said.

President Donald Trump said on Monday that the release of DeepSeek AI and its impact on the stock market should serve as a “wakeup call” for U.S. developers to come up with cheaper but reliable AI models.

“We always have the ideas. We’re always first. So I would say that’s a positive that could be very much a positive development,” Trump told Republican colleagues in Florida. “So instead of spending billions and billions, you‘ll spend less, and you’ll come up with, hopefully, the same solution.”

According to an analysis by The Epoch Times, DeepSeek skews heavily in favor of the Chinese Communist Party (CCP). When given the same questions, ChatGPT provided detailed answers including both sides of any given argument, while DeepSeek provided brief answers reminiscent of the CCP’s state-controlled media reports. The Chinese-trained AI model outright refused to answer questions about human rights.

The DeepSeek app reported a “major outage” affecting its application programming interface and user logins after its launch. DeepSeek said on Jan. 27 that it was temporarily limiting user registrations due to “large-scale malicious attacks” targeting its services, without elaborating further.

The Epoch Times has reached out to Graphika regarding its report but did not hear back by publication time.

Lily Zhou and Reuters contributed to this report.