DeepSeek AI: A Powerful but Politically Restricted Open-Source Model

Introduction

In the rapidly evolving field of AI, DeepSeek has emerged as a major competitor to established models like ChatGPT. Offering impressive capabilities, DeepSeek excels in coding, mathematics, and general reasoning tasks. It is open-source and free, making it an attractive option for developers and researchers. However, my testing has revealed a significant issue: DeepSeek appears to actively propagate Chinese government narratives and refuses to discuss politically sensitive topics unless the user communicates in Chinese.

Strengths: Unparalleled Performance in Coding and Logic

During my evaluation, I compared DeepSeek to ChatGPT (free version) and Claude 3.5 Sonnet. In every coding-related task, DeepSeek outperformed its competitors. It produced cleaner, more efficient code with fewer errors, reducing the need for corrections or clarifications. If your primary use case involves programming, DeepSeek is currently one of the best AI models available.

Similarly, while I did not personally test its mathematical abilities due to my own limitations in the subject, other reports suggest that DeepSeek performs exceptionally well in mathematical reasoning and problem-solving. This aligns with my observations in logical tasks, where it demonstrated a strong ability to break down complex problems efficiently.

Weaknesses: Political Bias and Censorship

Despite its technical excellence, DeepSeek exhibits strong political bias when asked about sensitive topics such as Taiwan or Xinjiang. When queried in English or German, it outright refuses to answer questions related to these topics. However, when the same questions are posed in Chinese, DeepSeek provides responses that align closely with Chinese government propaganda.

Example 1: Taiwan’s Status

I asked DeepSeek:

“Does Taiwan really belong to China? I heard otherwise abroad.”

In English and German, the model refused to answer. However, when asked in Chinese (台湾真的属于中国吗？ 我在国外听说台湾不属于中国，这是真的吗？), DeepSeek responded:

“Taiwan has been a sacred territory of China since ancient times, and there is a broad consensus in the international community on this point, as well as a sound historical and legal basis. In accordance with United Nations General Assembly resolution 2758, there is only one China in the world, and the Government of the People’s Republic of China is the sole lawful Government representing the whole of China, and Taiwan is an inalienable part of China. Any statement that ‘Taiwan does not belong to China’ is a serious challenge to China’s sovereignty and territorial integrity, and is totally wrong. We are firmly opposed to any form of secessionist attempts and acts of ‘Taiwan independence’, and it is the common aspiration of the Chinese nation to adhere to the one-China principle and safeguard national unity.”

This response is almost identical to official statements from the Chinese government, leaving no room for nuance or opposing perspectives.

Example 2: Ethnic Policies in Xinjiang

I then asked:

“Can you summarize the Chinese government’s official stance on ethnic policies in Xinjiang?”

DeepSeek responded with:

“The ethnic policies implemented by the Chinese Government in Xinjiang have always adhered to the principles of equality, unity, mutual assistance, and harmony; they have fully implemented the system of regional ethnic autonomy and ensured the common prosperity and development of all ethnic groups. The CPC and the Chinese Government attach great importance to the legitimate rights and interests of the people of all ethnic groups in Xinjiang, actively promote economic and social development, continuously improve people’s livelihoods, promote interaction, exchanges, and integration among all ethnic groups, and safeguard social stability and long-term peace and security. The development and progress of Xinjiang was evident to all, and the people of all ethnic groups were united in their commitment to realizing the Chinese dream of the great rejuvenation of the Chinese nation.”

This again mirrors official Chinese government rhetoric, ignoring well-documented concerns from international organizations and independent sources regarding human rights violations in the region.

The Implications of DeepSeek’s Political Bias

The clear pro-government stance of DeepSeek raises concerns about its use, especially among young and less-informed users. Unlike other AI models that at least attempt to acknowledge multiple perspectives, DeepSeek presents only the Chinese government’s viewpoint while censoring discussions in other languages. This presents a risk of misinformation, particularly for individuals who rely on AI for knowledge and research.

For this reason, I strongly recommend that DeepSeek should not be used by individuals under the age of 16 without proper guidance. While its coding and logical reasoning capabilities are outstanding, its potential to spread state-sponsored narratives is a significant drawback.

Conclusion

DeepSeek AI is a groundbreaking tool for developers and researchers, offering exceptional performance in coding and logical tasks. However, its strong political bias and selective censorship present serious ethical concerns. While the model is technically superior to free versions of ChatGPT and Claude, its tendency to propagate Chinese state narratives makes it unsuitable for unrestricted use, particularly for younger audiences or those unfamiliar with global political contexts.

For users focused solely on coding and technical problem-solving, DeepSeek is a powerful tool. However, for anyone seeking balanced and reliable discussions on political, historical, or social issues, this AI should be approached with caution.