According to the online analysis firm Graphika, accounts linked to the Chinese government significantly promoted the debut of the AI models from Chinese startup DeepSeek last week, shortly before a downturn in U.S. tech stocks.

These accounts, including those belonging to Chinese diplomats, embassies, and state-run media, amplified media narratives around the launch, suggesting that DeepSeek posed a challenge to America’s leadership in the AI field, as revealed in a Graphika report shared with Reuters on Thursday.

The messaging strategy was executed across various platforms such as Elon Musk’s X, as well as Meta Platforms’ Facebook and Instagram, and Chinese networks like Toutiao and Weibo, stated Graphika.

“This activity illustrates how China can rapidly mobilize numerous participants to propagate narratives that portray Beijing as excelling beyond the U.S. in critical areas of global competition, particularly in advancing and deploying cutting-edge AI technologies,” said Graphika’s Chief Intelligence Officer, Jack Stubbs, to Reuters.

“We have routinely observed both explicit and subtle engagements by Chinese state-affiliated individuals among the initial groups harnessing AI to enhance their influence in the information landscape.”

Graphika also noted a video featuring pro-China and anti-Western viewpoints on a YouTube channel, which exhibited similarities to Shadow Play, a coordinated influence effort identified by the Australian Strategic Policy Institute in 2023, involving at least 30 different YouTube channels.

Responses from YouTube’s parent company, Alphabet; Meta; X; and the Chinese embassy in Washington, D.C., were not available immediately regarding the report.

The analysis indicated a brief surge in conversations comparing DeepSeek’s developments to OpenAI’s ChatGPT on X right after DeepSeek’s model release on January 20, followed by a more substantial increase that began on Friday and continued through the weekend.

By Monday, DeepSeek’s free AI assistant had surpassed ChatGPT in downloads from Apple’s app store, coinciding with a sell-off of U.S. tech stocks that resulted in a staggering $593 billion loss in market value for the chipmaker Nvidia, marking the largest one-day loss for any company on Wall Street.

Nvidia chose not to comment on the Graphika report.

DeepSeek’s researchers assert that they created components of their AI technology at significantly lower costs than their U.S. counterparts, raising concerns that American companies, which have invested billions into AI data centers, may soon face price competition from China.

Shares of Microsoft, a significant backer of OpenAI operating data centers for ChatGPT, dropped earlier this week as it announced cloud revenue growth that fell short of Wall Street expectations while continuing its substantial capital expenditures.

Both Microsoft and Meta have committed to maintaining significant investments in AI for the foreseeable future.

In China, DeepSeek’s success is viewed as a symbol of the country’s resilience against U.S. efforts to restrict its technology sector through export limitations.

Meanwhile, in the U.S., there are accusations suggesting DeepSeek may have unlawfully accessed technologies from OpenAI and other leading firms, though no evidence has yet been proven to support this claim.

Additionally, the U.S. Commerce Department is investigating whether DeepSeek has utilized U.S. chips that are not permitted for export to China, according to a source familiar with the situation.

Reporting by Katie Paul in New York and Stephen Nellis in San Francisco; Editing by Himani Sarkar