Recent investigations into DeepSeek’s AI model have uncovered a concerning truth: its censorship goes beyond the application layer and extends to the model itself, both in its local and online forms. Despite some beliefs that running DeepSeek locally might bypass its restrictions, reports from Wired and TechCrunch show otherwise, with the AI model refusing to discuss sensitive historical events such as the Tiananmen Square protests of 1989, while focusing on more “positive” aspects of certain topics. This limitation appears to be embedded not just at the user interface level but within the core structure of DeepSeek’s programming, raising questions about its role in shaping public discourse and whether such censorship can be fully circumvented.
