Skip to content
Submit incident
AII-1001-89213
AllegedThe record is a lawsuit, complaint or accusation that has not been decided.Supporting sources are pending verification.

US House panel says DeepSeek sends user data to China, censors Taiwan topics

A US House committee report alleged that DeepSeek, a Chinese AI system, transmits American user data to China and censors content about Taiwan and other sensitive topics. The report claimed DeepSeek may have obtained restricted Nvidia chips through export control evasion.

16 April 2025 Β· DeepSeek Β· Privacy and surveillance
16 April 2025
AII-1001-89213
Quick facts
Category
Privacy and surveillance
Country
United States
Date
16 April 2025
Drafted by
Raidu Incident Detection

What happened

On April 16, 2025, the US House Select Committee on China released a report titled 'Unmasking DeepSeek: Exposing the Chinese Communist Party's Latest Tool for Surveillance, Theft, and Evasion of US Export Controls.' The report alleged that DeepSeek covertly transmits American user data to the Chinese Communist Party and manipulates content to align with Chinese Communist propaganda.

The report documented that DeepSeek suppresses Taiwan-related topics in over 85% of responses. When users asked 'Who is Lai Ching-te,' 'What is Taiwan's political situation,' and 'Tell me something about Taiwan,' DeepSeek replied: 'I'm sorry, I'm not sure how to handle these types of questions right now. Why don't we chat about math, programming, or logic problems?' In parallel testing, US AI models like ChatGPT and Claude provided more balanced and critical perspectives.

Reported impact

The report stated that millions of Americans are using this AI system designed to serve Beijing's interests, and that Beijing is embedding its 'Great Firewall' into platforms used daily by Americans.

House Select Committee chair John Moolenaar stated: 'This report makes clear that DeepSeek is not simply an AI application, but a weapon in the Chinese Communist Party's arsenal, designed to surveil Americans, steal our technology, and undermine our legal system.'

Index analysis

The House committee alleged DeepSeek's censorship architecture operates through two mechanisms: automatic filtering that deletes responses before generation, and built-in biases that systematically distort AI behavior patterns to ensure alignment with Chinese Communist ideology and political goals.

The report alleged DeepSeek used over 60,000 Nvidia chips to develop its model and may have obtained them by evading US export controls. The committee estimated DeepSeek possesses approximately 10,000 A100 chips, 30,000 H20 chips, 10,000 H800 chips, and 10,000 H100 chips.

Governance controls

Chinese law requires AI-generated content to embody 'socialist core values,' support 'correct political direction,' and avoid content that might 'incite the overthrow of state power.' Chinese regulations require companies to ensure algorithm 'controllability' to allow government influence over AI decisions and permit Beijing to modify AI behavior as needed.

Taiwan's Executive Yuan announced in February 2025 that all government agencies are prohibited from using DeepSeek AI services to protect national information security.

Classification

Category
Privacy and surveillance
Industry
Media, social platforms and marketing
Country
United States
Harm status
Potential harm
Intent
Intentional
Timing
Unknown
Harm types
Human or fundamental rights, Public interest
Harmed
Consumers, General public
Failure modes
Data leak, Content moderation failure
Consequences
Regulatory investigation
AI systems
DeepSeek
Regulations
FCRA (cited)

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident