American AI chatbots, and Gemini especially, collect more user data than Chinese alternatives

New research challenges the prevailing narrative about Chinese AI models posing the greatest privacy risks, revealing instead that popular American chatbots often collect more personal data. This counterintuitive finding comes amid heated international debate over AI ethics and regulation, as users worldwide increasingly rely on AI assistants while governments struggle to establish appropriate privacy safeguards across different jurisdictions.

The big picture: Despite widespread suspicion of Chinese-developed DeepSeek R1, research from VPN provider Surfshark finds it ranks only fifth for data collection among popular AI chatbots.

• American-developed Google Gemini tops the list as the most data-intensive AI chatbot, collecting 22 out of 35 possible user data types.
• The study analyzed privacy details for the ten most popular chatbots on Apple’s App Store, including ChatGPT, Gemini, Copilot, Perplexity, DeepSeek, Grok, Jasper, Poe, Claude, and Pi.

Key details: DeepSeek’s controversial Chinese origins have fueled privacy concerns, prompting bans by government and private organizations in the US and elsewhere.

• DeepSeek’s flagship open-source AI model debuted in January, attracting approximately 12 million users within just two days of its launch.
• The company’s privacy policy acknowledges storing personal information on secure servers located in China, which has intensified concerns among Western users.

By the numbers: Google Gemini collects the most extensive range of sensitive user information among popular AI chatbots.

• Gemini harvests highly sensitive data including precise location, user content, contacts lists, and browsing history.
• Only three chatbots—Gemini, Copilot, and Perplexity—collect precise location data from users.
• Approximately 30% of chatbots share sensitive user data with third parties, and the same percentage actively track user data.

For comparison: DeepSeek collects an average of 11 unique data types, focusing primarily on contact information, user content, and diagnostics.

• ChatGPT, developed by US-based OpenAI, collects 10 unique data types, including contact information, user content, identifiers, usage data, and diagnostics.
• This places both popular AI assistants in the middle range for data collection practices among the chatbots analyzed.

Why it matters: The findings highlight how geopolitical tensions may distort public perception of privacy risks, with users potentially overlooking excessive data collection by domestic companies while focusing concerns on foreign alternatives.

• The research emerges amid an escalating AI arms race between the US and China, which continues to raise profound privacy, security, and ethical questions on both sides.