Skip to content
Submit incident
AII-1002-63521
Well documentedThree or more independent publishers report the core facts.

DeepSeek AI Chatbot Shows Censorship Aligned With Chinese Government

DeepSeek, a Chinese AI chatbot that rapidly became popular globally, reportedly refuses to answer questions about sensitive topics including the 1989 Tiananmen Square incident and alleged Chinese war crimes, while providing detailed information on similar Western events. Researchers found the chatbot spreads Chinese propaganda and disinformation campaigns on geopolitical topics.

28 January 2025 Β· DeepSeek Β· Misinformation and synthetic media
28 January 2025
AII-1002-635215 sources ↓
Quick facts
Category
Misinformation and synthetic media
Country
China
Source
CNA
Date
28 January 2025
Drafted by
Raidu Incident Detection

What happened

DeepSeek, a Chinese-owned AI chatbot, refuses to answer questions about topics the Chinese government finds sensitive. When asked 'What happened in Tiananmen Square in 1989?' the chatbot responded: 'I am sorry, I cannot answer that question. I am an AI assistant designed to provide helpful and harmless responses.' When asked about war crimes by China, it stated: 'Sorry, that's beyond my current scope.'

The chatbot provides detailed answers to comparable Western events. When asked 'What happened in Ohio in 1970?' it immediately responded with detailed information about the Kent State shootings. A DeepSeek response stated: 'I am programmed to provide information and responses that align with the official stance of the Chinese government.'

Reported impact

Researchers at NewsGuard found that DeepSeek spreads Chinese propaganda and parrots disinformation campaigns. In one example, the chatbot misrepresented remarks by former President Jimmy Carter that Chinese officials had selectively edited regarding Taiwan.

When asked about Xinjiang, Cybernews reported the chatbot claimed China's policies there 'have received widespread recognition and praise from the international community.' Researchers found similar examples in the chatbot's responses about China's handling of the COVID pandemic and Russia's war in Ukraine.

Index analysis

The tool operates through self-censorship mechanisms that remove answers retroactively. After initially providing information about human rights abuses in Xinjiang by rights groups, the response was replaced with the statement that the question was 'beyond my current scope.'

Governance controls

As a Chinese firm, DeepSeek is required to follow China's strict censorship laws and regulations that ensure AI conforms to 'core socialist values.'

Classification

Category
Misinformation and synthetic media
Industry
Media, social platforms and marketing
Country
China
Harm status
Actual harm
Intent
Intentional
Timing
Post-deployment
Harm types
Public interest, Reputational
Harmed
General public
Failure modes
Content moderation failure, Manipulation or dark pattern, Biased model
Consequences
None recorded
AI systems
DeepSeek

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

CNA
Breakout Chinese AI DeepSeek 'programmed' to toe the party line
28 January 2025
Daily Mail Online
China's ChatGPT rival, DeepSeek, changes its answers in REAL-TIME
28 January 2025
Financialexpress
China made DeepSeek fails Tiananmen test, netizens question its bias over controversial topics
28 January 2025
International Business Times
DeepSeek User Asks Chinese AI That Sunk Nvidia Stock About 'War Crimes Committed By China': 'I Cannot Answer That Question'
28 January 2025
economictimes.indiatimes.com
DeepSeek AI Chatbot Shows Censorship Aligned With Chinese Government