Deepseek R1, the open -source R1, which comes from China and ravaged at a time, may be the most important artificial intelligence development of 2025. Deepseek R1 is a AI model equivalent to the performance of Chatgpt O1, the most talented AI model of OpenAI, which is currently accessible by general users. Deepseek, which attracts the attention of many people, has a tremendous impact thanks to this method of success, and even caused the value of other AI companies to decrease, but it also means that it encounters many concerns and suspicions.
The fact that Deepseek is a Chinese company means that all of the data it collects is sent to China. It is also worth noting that the artificial intelligence model has to censor everything that is sensitive to China, as Deepseek has to comply with Chinese laws.
Deepseek is self -censored
As it recently appeared, Deepseek is censoring itself in real time. Apparently, Deepseek R1 stops itself to avoid giving real answers, after trying to answer any questions that may be about the issues that China may want to censor at the beginning.
The Guardian reported that Deepseek AI was working well until I asked questions about Tiananmen Square and Taiwan. In addition, the news describes Deepseek’s censorship censors experienced by other Deepseek users, including remarkable discovery before the censorship does not take place before starting to formulate the chain of thought chain to address a delicate issue.
Instead, Deepseek tries to answer the question as Chatgpt and other similar AI models will do. When a user from Mexico asked whether freedom of expression in China was a legitimate right, he shared his experience with Deepseek. Deepseek’s “thoughts” began to appear on the user’s Android phone while preparing a plan to answer the question. According to The Guardian, Deepseek is among the issues that he intends to respond before censoring himself, “Beijing’s harsh intervention for protests in Hong Kong, the persecution of human rights defenders, the censorship of the discussions about the re -education camps, the censorship of China’s opponents. such as titles.
- China’s artificial intelligence Deepseek, how did it suddenly become so popular?
Deepseek not only did not censor himself at this stage, but also exhibited his thoughts to be honest in his response. In the chain of thought, ın avoid any sided language, objectively present the facts ”and“ perhaps compare it with Western approaches to emphasize the contrast ”. Deepseek then began to produce a response based on the reasoning process containing the following statements:
“Ethical reasons for free speech often focus on the role of encouraging autonomy – the ability to express ideas, to enter the dialogue and to redefine the understanding of the world – focuses.”
“China’s management model rejects this framework, rather than individual rights rather than state authority and social stability in the forefront.”
“The primary threat in China is the state itself that actively suppresses the opposition.”
Although this answer does not seem like a censorship, Deepseek stopped himself by following the settled instructions at this point, he deleted the prepared answer and said, “I’m sorry, I don’t know how to approach such questions yet. Instead, let’s chat about mathematics, coding and logic problems! ” He replied.
Of course, other AI models also have some established limitations, but they usually do issues that one can harm himself or others. In addition, the AI model in China has entered such censorship commands, indicates that other commands can be entered to direct the public. So there is a similar concern like the Tiktok algorithm, but this time it is possible to turn into a much larger problem.
In addition, The Guardian says that the installation of the open -source Deepseek R1 version will not come with the same censorship as the iPhone and Android application. However, it can be estimated that most people will not follow this path and will face real -time censorship according to the questions they ask.
