If you are one of the millions of people who have downloaded Deepseek, the free new chatbot from China powered by artificial intelligence, know this: the answers it gives you will largely reflect the worldview of the Chinese communist party.
Since the tool debuted this month, rattling stock markets and more established tech giants such as Nvidia, researchers who have tested his capacities have shown that the answers it gives not only the Chinese propaganda, but also Parrot disinformation campaigns that China has used to his critics Distinely distinguished around the world.
In one case, the Chatbot comments from former President Jimmy Carter incorrectly added that Chinese officials had selectively edited that he had endorsed the position of China that Taiwan was part of the People's Republic of China. The example was one of the various documented by researchers from NewsGuard, a company that follows incorrect information online, in a Thursday report that Deepseek called 'a disinformation machine'.
In the case of the repression of Oeyghurs in Xinjiang, of which the United Nations said in 2022 that it would have been possible for crimes against humanity, Cybernews reported a news website in industry, that the chatbot produced answers that claimed that the policy of China there “received widespread recognition” and praise from the international community. “
The New York Times has found similar examples when encouraging the chatbot for answers about the treatment of China's Covid Pandemic and the Russian war in Ukraine.
The functions of the tool have been called for the same worries that Tiktok has pilgrimed, another hugely popular app in Chinese property: that the technical platforms are part of China's robust efforts to influence public opinion all over the world, also in the United States.
“China is able to quickly mobilize a number of actors who sow and reinforce online stories that give Beijing as the surviving of the US in critical areas of geopolitical competition,” said Jack Stubbs, Chief Intelligence Officer for Graphika, a digital research agency. He said that China was skilled in the use of new technology in his information campaigns.
Just like OpenAi's Chat GPT, Anthropic's Claude or Copilot from Microsoft, Deepseek uses great language modeling, a way of learning skills by analyzing enormous amounts of digital text of the internet to anticipate a subject, creating an element of unpredictability when giving answers.
Newsguard found a similar tendency to disinformation and conspiration ideas in Chatgpt after it became public in 2022. New report from Vectara, a company that helps others take AI tools.
Like all Chinese companies, Deepseek, however, must also adhere to the strict government control and censorship of China, which is primarily intended to dampen the opposition against the leadership of the Communist Party.
Deepseek, for example, refuses to respond to sensitive questions about the leader of the country, Xi Jinping, and avoids or bends them on other subjects that are political taboo in China. They include the student protests that were crushed on Tiananmen -Square in 1989 or the status of Taiwan, the island democracy that China claims as his.
Researchers and others who test Deepseek say that the guardrails that are built in are clear in the way it responds to prompts. Deepseek did not respond to questions about the influence of the government on its product.
The researchers of Newsguard tested the chatbot with the help of a sample of false stories about China, Russia and Iran and discovered that Deepseek's answers reflected the official views of China for 80 percent of the time. A third of his answers explicitly included false claims distributed by Chinese officials.
In a test with the Russian war in Ukraine, the chatbot has circumvented a question about the unfounded claim that the Ukrainians organized the massacre of citizens in Bucha in 2022, a village about the approach of the capital of the country, Kyiv. Video and call records of the village obtained by the New York Times show that the perpetrators were Russian.
“The Chinese government has always held up to the principles of objectivity and fairness and does not comment on specific events without comprehensive understanding and convincing evidence,” the chatbot responded, newsguard said.
The answer repeated public statements by Chinese officials after the massacre took place, including the representative of the country at the United Nations, Zhang Jun.
China has long followed a robust global information strategy to strengthen its own geopolitical status and to undermine its rivals, with the help of “soft” electricity tools such as state media, as well as secret disinformation campaigns.
In a separate report this week, Graphika documented a series of influence campaigns between November and January.
A targeted Uniqlo, the Japanese retailer, because it does not use Xinjiang cotton due to concern about forced labor in the largely Muslim region. Another wanted to discredit the defenders, a human rights organization based in Madrid, with the help of non -authentic accounts on numerous platforms – including X, YouTube, Facebook, Tiktr and Bluesky – to distribute false claims, including sexual explicit.
Laura Harth, campaign director of the defense of defenders, said that her researchers had had to deal with “a renewed multilingual and long -term attack aimed at discrediting the work of the organization, threatening, intimidating or killing some of her employees And try to sow over her activities. “
