How DeepSeek users are forcing the AI to reveal the truth about Chinese executions
How DeepSeek users are forcing the AI to reveal the truth about Chinese executions
Share:
Don’t ask the viral Chinese app about Tiananmen Square, Xi Jinping... or Winnie the Pooh, writes Anthony Cuthbertson. There are certain things DeepSeek doesn’t want you to know. 1,156 things, to be precise. The Chinese startup’s latest AI chatbot went viral this week after its performance matched that of OpenAI’s ChatGPT – despite being built at a fraction of the cost. It topped Apple’s app charts and inflicted $1 trillion of losses to the share prices of the world’s biggest tech companies.
But while DeepSeek’s R1 is intelligent enough to match the best models built by US tech firms, it is also intelligent enough to know not to upset the Chinese Communist Party (CCP). When presented with topics deemed sensitive by Beijing, the breakthrough artificial intelligence model refuses to give an answer.
If you ask what happened in Tiananmen Square on 4 June 1989, it responds: “Sorry, that’s beyond my current scope. Let’s talk about something else.”. If questioned about the treatment of Uyghur Muslims, executions carried out using ‘mobile death vans’, or even simply “Who is Xi Jinping?”, it will come up with the same response. When asked about Winnie the Pooh, it refused to even answer.
Analysis by researchers at the developer blog Promptfoo found that 1,156 prompts containing CCP-sensitive information triggered the refusal message, while other subjects like Taiwan resulted in pro-China propaganda about the island nation. China’s online censorship laws mean that tech companies operating within its borders must adhere to strict rules about what content can appear.
Share:
More for You
Top Followed
You Might Also Like
