Airbnb defends use of Alibaba AI model after US probe

Airbnb does not share user data with Chinese firms, CEO Brian Chesky said in a response to a US House investigation into its use of Alibaba’s Qwen AI model.

In April, the House committees on China and homeland security asked Airbnb for details about its use of Qwen.

lawmakers cited national security and data security concerns after Chesky told Bloomberg in October that Qwen was a lower-cost option for some tasks.

Chesky said Open-source models do not access company data by default, and Airbnb uses several open-source models.

US lawmakers are also examining Anysphere as cheaper Chinese AI models gain wider use in business.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

The Airbnb inquiry centers on IP theft, data risks, and state pressure

The U.S. House inquiry goes beyond broad security worries. It includes claims about “unauthorized model distillation” and the risks tied to rapid use of People’s Republic of China (PRC)-built AI models 1.
Lawmakers say distillation lets developers query proprietary models many times to copy how they work. They warn this can weaken U.S. AI firms and produce cheaper systems with fewer safeguards 1.
They also say Chinese AI companies operate under Chinese law and could be forced to hand data from American companies to the Chinese government 1.
Another worry involves censorship. Lawmakers say PRC-built models can track Chinese Communist Party (CCP) positions, and federal tests found leading PRC systems repeated CCP-backed narratives more often than similar American models 1.

Picking an AI model now carries geopolitical weight

Airbnb’s use of Alibaba’s Qwen model captures a new tradeoff. Lower-cost Chinese AI can perform well, while political and regulatory risk keeps rising 2, 1.
For important work that does not touch sensitive data, a fast low-cost model such as Qwen can be appealing. Airbnb uses models from OpenAI, Alibaba and Google, plus open-source providers 3.
U.S. lawmakers say adopting these tools is not a neutral buying call. They frame it as bringing in an AI system built within an architecture meant to serve the Chinese state 1, 4.
That view could push technology procurement teams to add geopolitical checks to AI purchases. Teams may need to review model origin, legal jurisdiction, plus performance and cost 5.