China is openly sharing its AI — should the West be locking it out?

The launch of Chinese firm DeepSeek’s open-source AI model has sparked intense debate about AI development and national security, particularly regarding U.S.-China technological competition.

Key developments: DeepSeek’s latest AI model, which rivals U.S. closed-source systems, has become globally popular while simultaneously promoting Chinese Communist Party (CCP) viewpoints.

• The model has become the most downloaded app globally despite promoting One China policy and avoiding discussion of sensitive topics like Uyghur genocide
• The software is freely available and modifiable, allowing developers to study and alter its behavior
• Within days of release, developers created over 500 variations of the model, generating five times more downloads than the original

Legislative response: U.S. lawmakers have proposed aggressive measures to restrict AI technology exchange with China.

• Senator Josh Hawley introduced legislation to ban AI technology import and export with China, carrying penalties of up to 20 years imprisonment
• The bill would prohibit research collaborations with Chinese universities
• The proposed restrictions would cover chips, data, research, software, and model parameters

Technical implications: DeepSeek’s open-source approach offers both opportunities and challenges for the global AI ecosystem.

• The model matches U.S. competitors in capability while being more efficient and cost-effective
• AI search engine Perplexity has modified the model to bypass original censorship constraints
• Unlike major U.S. companies like OpenAI and Google, which keep their models closed and paywalled, DeepSeek’s open approach allows for transparency and modification

Industry perspective: Critics argue that U.S. companies have focused too heavily on content moderation at the expense of competition.

• David Sacks, an AI and crypto expert, suggests U.S. firms “wasted time” on diversity and inclusion initiatives rather than advancing core technology
• U.S. policymakers face a contradiction: criticizing both Chinese ideology in open models and U.S. firms’ closed-model censorship
• There’s growing concern that China-regulated models could become the default foundation for future AI applications globally

Strategic considerations: The situation presents a complex challenge for U.S. policy and technological leadership.

• Open-source technology allows for inspection and modification but can be misused
• Restricting open-source development could inadvertently promote reliance on Chinese technology
• U.S. remains crucial to the global AI supply chain through chips, research, data, and capital

Looking ahead: The AI crossroads: Rather than implementing restrictive measures, the U.S. faces a strategic imperative to maintain leadership in open-source AI development while balancing security concerns. The outcome of this technological competition could determine whether future global AI systems align more closely with U.S. or Chinese interests, making it crucial to find a balanced approach that promotes innovation while protecting national interests.