China has officially begun to compete in the AI landscape with the release of DeepSeek, a startup that has rapidly positioned itself as a formidable competitor in the AI sector. Liang Wenfeng started developing it in 2023 and has been able to create the R1 model, an AI system that rivals leading models like OpenAI’s ChatGPT but at a fraction of the development cost.
R1 uses “pure reinforcement learning” to improve its capabilities in reasoning, mathematics, and coding tasks. Additionally, DeepSeek has adopted an open-source approach, making its AI algorithms and training methodologies publicly accessible.
Following the release of the R1 model, DeepSeek’s application quickly became the most downloaded free app on the iOS App Store in the United States, surpassing established competitors like ChatGPT. This popularity contributed to a drop in stock prices for several major U.S. tech companies, including Nvidia, Microsoft, and Alphabet, totaling approximately $1 trillion in market value.
AI explained: What artificial intelligence is and how it can work for you
China has made this significant progress in AI development even with U.S. export controls on advanced Nvidia chips and the open-source nature of DeepSeek, potentially democratizing access to advanced AI technologies.
Other Chinese tech giants have also intensified their efforts in AI development since DeepSeek’s success. Alibaba, for instance, has introduced Qwen 2.5 Max, an AI model it claims outperforms both DeepSeek’s and OpenAI.
However, while DeepSeek is extremely popular in the U.S., insiders warn that the technology is raising concerns regarding data privacy and security. Much like the problem with TikTok, applications hosted on servers in China could pose risks related to data protection and potential misuse, given the differences in data privacy regulations between China and Western countries.
There are also potentially a few issues with the emerging technology. For instance, Open AI, the creators of ChatGPT, have accused DeepSeek of copying technology through a process called “distilling,” which involves extracting knowledge from existing models. OpenAI and its major investor, Microsoft, are investigating potential data theft and copyright infringement by DeepSeek.
Another potential problem is that the model employs censorship mechanisms, particularly on topics deemed politically sensitive by the Chinese government, and won’t answer any questions about human rights issues, protests, and many other issues that occur in the country. This built-in censorship has raised concerns about the limitations on access to information and the potential influence of political considerations on AI responses.
An audit conducted by NewsGuard found that DeepSeek’s chatbot achieved only 17% accuracy, trailing behind Western rivals like OpenAI’s ChatGPT and Google’s Gemini.
Follow GeekSided to stay up to date with emerging AI technology.