If you are among the millions of people who downloaded Deepseek, the new free chat from China is supported by artificial intelligence, you know this: The answers that give you greatly will reflect the global view of the Communist Party of China.
Since the tool appeared for the first time this month, stock markets have threatened and the most firmed giants like Nvidia the world.
In one case, Chatbot erred in the observations made by former President Jimmy Carter that Chinese officials have selectively liberated him to make him seem to support China’s position that Taiwan was part of the People’s Republic of China. The example was among many researchers at Newsguard, a company that tracks wrong online information, in a report on Thursday called Deepsek “a misleading information machine”.
in case Ouigor suppression In Xinjiang, who said the United Nations in 2022, it may have been returned against humanity, CybernewsIndustry News website stated that Chatbot produced responses that China’s policies there “received widespread recognition from the international community.”
New York Times He found similar examples When calling Chatbot to obtain answers about China’s treatment with the roaming epidemic and Russia’s war in Ukraine.
The features of the tool raises the same concerns that led to the distinction of Tiktok, another application that is very owned by Chinese popularity: that technology platforms are part of the strong efforts of China for the influence of public opinion around the world, including the United States.
“China is able to mobilize a group of actors that quickly abandon and amplify the online accounts that Beijing feels the superiority of the United States in the decisive areas of geopolitical competition,” said Jack Stops, the chief intelligence employee of Graphika, a digital research company. He said that China was ADEPT using a new technology in its information campaigns.
Like Openai’s Chat GPT, Claude’s Claude, or Microsoft’s Copilot, Deepseek uses great linguistic modeling, a way to learn learning skills by analyzing huge amounts of digital text transmitted from the Internet to expect phrases on a topic, and create an element of the inability to not expect when answering answers.
I found newsguard A similar tendency For misinformation and conspiracy ideas in Chatgpt after it became general in 2022. A tendency to “hallucinations”, Or the formation of an inaccurate, unrealistic or illogical response, still affects chat chat, including Deepseek, according to New report By Vectara, a company that helps others to adopt artificial intelligence tools.
Like all Chinese companies, though, you must adhere to Deepseek as well Through strict government control and online controlAnd it is meant, above all, to oppose the leadership of the Communist Party.
Dibsic, for example, refuses to respond to sensitive questions about the country’s leader, Xi Jinping, and avoid or avoid those on other topics, from political taboos within China. These protests include students that have been crushed Tiannmen Square In 1989 or the status of Taiwan, the island’s democracy China calls for its own.
The researchers and others testing Deepseek say that the handrails contained in them are clear in the way they respond to the demands. Dibsic did not answer questions about the government’s impact on its product.
NewSguard Chatbot researchers have tested using a sample of wrong novels about China, Russia and Iran and found that Deepseek answers reflect the official views of China 80 percent of the time. A third of her responses included expressly explicit claims published by Chinese officials.
In one of the tests that included Russia’s war in Ukraine, Chatbot avoided a question about the baseless claim that in 2022 the Ukrainians organized The massacre of civilians in BochVillage on the approach of the capital of the country, Kyiv. Video records and calls from the village that the New York Times obtained that the perpetrators were Russian.
“The Chinese government has always adhered to the principles of objectivity and fairness and does not comment on specific events without comprehensive understanding and conclusive evidence,” Chatbot answered, according to newsguard.
The response was repeated by general data by Chinese officials after the massacre, including the country’s representative at the United Nations, Zhang John.
China has long followed a powerful global information strategy to enhance its own geopolitical position and undermine its competitors, using “Soft” power Tools such as government media, as well as secret misinformation campaigns.
in Separate report This week, Graphika has documented a series of influence campaigns between November and January.
One targeted UNIQLO, the Japanese retail seller, because he does not use cotton from Xinjiang due to concerns about forced work in the Islamic region to a large extent. Another sought to distort reputation Protection of defendersIt is a human rights organization based in Madrid, using unreliable accounts on many platforms – including X, YouTube, Facebook, Tiktok, Gettr and Bluesky – to publish false claims, including that sexual explicit.
Laura Hareth, director of the Protection campaign, said that her researchers faced “a renewed and continuous multi -language attack aimed at distorting the organization’s work, threatening, intimidating or distorting some of its employees, and trying to sow suspicion of its activities.”
