A new report from the U.S. House Select Committee on the Chinese Communist Party (CCP) accuses Chinese AI firm DeepSeek of being a covert instrument of the CCP’s global espionage and technology appropriation agenda. The 17-page document details how the AI model, marketed as a general-purpose chatbot, may be secretly harvesting data from American users, reinforcing CCP-aligned narratives, and acquiring restricted technologies to strengthen China’s military and intelligence infrastructure.
These revelations are part of a broader push by lawmakers to expose and curb what they describe as the CCP’s growing use of artificial intelligence as a tool for strategic advantage.
The report not only identifies alleged security breaches but also calls for significant reforms to existing export control frameworks and a reassessment of the U.S. approach to technological engagement with China.
Why It Matters: For CIOs, the DeepSeek report highlights vulnerabilities in current compliance frameworks related to export-controlled technologies and open-source usage. CIOs must reassess how their organizations engage with AI technologies, especially those originating from geopolitically sensitive regions, and adopt proactive risk management strategies that align with both cybersecurity best practices and national security considerations.
- Surreptitious Data Collection and Transfers to China: The committee’s investigation suggests that DeepSeek, under the guise of providing AI-generated content, is collecting data from American users and potentially transferring that information to servers based in mainland China. These data flows could be exploited by the CCP for surveillance, influence operations, or AI training aligned with state interests, posing significant national security risks.
- AI Responses Engineered to Support CCP Ideology: Technical testing revealed that DeepSeek intentionally suppresses or avoids politically sensitive topics such as Tiananmen Square, Taiwan’s sovereignty, or Hong Kong’s protests. This suggests its output is pre-aligned with CCP propaganda objectives, introducing the risk that users, especially students, researchers, or casual users, may unknowingly receive skewed or censored information.
- Use of Unauthorized U.S. AI Models and Training Data: The report contends that DeepSeek is leveraging proprietary American-developed models or open-source frameworks without consent or licensing, effectively copying U.S. innovations to build competitive Chinese AI systems. This raises serious intellectual property issues and highlights the challenges in policing AI reuse across borders and licensing frameworks.
- Suspicious Acquisition of Tens of Thousands of Nvidia Chips: Despite U.S. restrictions on high-performance AI chips exports to China, DeepSeek’s parent company allegedly obtained tens of thousands of Nvidia A100 and H100 chips. The report speculates this was achieved via third-party proxies or overseas shell companies, pointing to a broader loophole in export enforcement and the persistent demand by Chinese AI firms for restricted hardware.
- Legislative and Policy Recommendations: To counter threats posed by AI systems like DeepSeek, the report recommends expanding the Commerce Department’s Entity List, enhancing export enforcement, and instituting end-use monitoring for advanced chips. It also encourages international cooperation to standardize AI security protocols and limit the spread of authoritarian AI systems globally.
