DeepSeek accused of risking US security with data to China

A bipartisan House committee has accused AI startup DeepSeek of jeopardizing United States national security by allegedly collecting data from American users and sending it to China.

These claims were detailed in a report released on April 16, 2025 by the House Select Committee on China.

The report states that DeepSeek’s founder, Liang Wenfeng, and hedge fund High-Flyer Quant operate in an “integrated ecosystem” associated with Chinese state-backed organizations, including hardware distributors and Zhejiang Lab, a research institute.

It also alleges that DeepSeek’s AI model censors and modifies information to comply with Chinese law.

🔗 Source: Bloomberg

🧠 Food for thought

1️⃣ DeepSeek represents the latest chapter in a long-running US-China tech security saga

The House committee’s accusations against DeepSeek continue a pattern of escalating US scrutiny of Chinese technology firms that spans multiple administrations.

This follows earlier government actions like the 2019 blacklisting of Chinese supercomputing companies and Huawei, which similarly cited national security concerns 1.

The DeepSeek allegations echo previous cybersecurity incidents, including the 2018 indictment of Chinese hackers affiliated with China’s Ministry of State Security who were charged with stealing intellectual property from US companies 2.

Security experts have consistently raised concerns about data theft costing the US economy, with estimates suggesting technology theft from China costs over $57 billion annually 3.

The heightened attention to DeepSeek reflects the broader context where technology competition has become central to US-China strategic rivalry, particularly in transformative areas like AI.

2️⃣ DeepSeek’s cost claims challenge fundamental assumptions about AI development economics

DeepSeek’s reported $5.6 million model training cost represents a stark contrast to OpenAI’s estimated $3 billion annual training costs, potentially disrupting the economics of AI development 4.

This cost discrepancy has already triggered market reactions, including fluctuations in tech stocks, as investors reassess the barriers to entry for competitive AI development 5.

DeepSeek’s founder has emphasized optimizing existing technologies rather than simply replicating them, suggesting an alternative approach to AI innovation that prioritizes efficiency 5.