The Tamesis
SEOUL, SOUTH KOREA – JUNE 4: Airbnb co-founder and CEO Brian Chesky speaks during a press conference ahead of the SEVENTEEN Airbnb Experience celebrating the group’s 10th anniversary at Replace Hannam on June 4, 2025 in Seoul, South Korea. (Photo by Myunggu Han/Getty Images for Airbnb)
Getty Images of Airbnb
Airbnb CEO Brian Chesky defend His company used a Chinese artificial intelligence model this week, saying U.S. lawmakers concerned about Chinese companies’ access to U.S. data were “misunderstanding” the technology.
Chesky clarified on Bloomberg TV that Airbnb is not a customer of Alibaba or other Chinese companies interview. “We will not provide data to any Chinese company. They have no access to any data,” he said in an earlier interview with Bloomberg News, his first public response to a U.S. House investigation into Airbnb’s use of Alibaba’s Qwen large language model in its customer service chatbot.
“We mostly use various open source models, including the open source model in the United States. The open source model doesn’t have access to the data. It doesn’t work that way. I think people need to understand how these things work.”
The comments come as the race for dominance in artificial intelligence between the United States and China escalates.
Last month, the U.S. House of Representatives China and Homeland Security Committee sent a letter to Airbnb asking it to clarify its use of Chinese artificial intelligence models in an effort to investigate what they said was China’s activities to “accelerate its artificial intelligence capabilities by leveraging U.S. innovation.”
The panel leader cited an October interview with Bloomberg in which Chesky said his company preferred “fast and cheap.” Kuiwen In some cases. Committee members cited those comments and noted that Airbnb has “serious concerns about the national security and data security implications” for its “U.S. customers and the integrity of its systems.”
Committee Chairman John Moolenaar said semafor “The AI models used by these companies were trained by Chinese censorship and introduced hidden vulnerabilities that put U.S. data and businesses at risk.” This letter Going further, he argued that Alibaba’s Qwen was ideologically constrained, thereby introducing structural holes – something Chesky objects to.
These developments add to an interesting and increasingly crowded and thorny conversation around the tension between Silicon Valley pragmatism and Washington’s national security instincts, at a time when the gap is widening almost weekly.
In a direct sign of the pressure, U.S. Senators Jeanne Shaheen and Pete Ricketts introduced the bipartisan bill on Tuesday. American Technology Pathways Act Streamlining allied government procurement of U.S. cyber and digital technology, contrary to what the bill itself describes as foreign partners “increasingly turning to strategic competitors such as China to procure cyber and digital technology because these technologies are low-cost.”
If passed, the bill would create an Office of Technology Procurement led by the State Department and authorize $500 million over the next five years to fund the program.
“Our competition with China centers on our ability to develop and promote future technologies to our partners,” Shaheen said. “This legislation sends a message to the world: The United States will compete on technology and we can offer better deals.”
Why is this important?
An important clue surrounding this issue is the widespread availability of low-cost AI models from China that perform nearly as well as expensive products from top U.S. labs; arguably, it doesn’t make as much sense to use U.S. or other local technology if cheaper alternatives are available elsewhere.
Based on empirical research on 100 trillion tokens open routerChina’s global share of open source LL.M.s rose from a low base of 1.2% at the end of 2024 to nearly 30% within a few months of 2025. study Researchers from MIT and Hugging Face found that China’s open-weighted models accounted for 17.1% of global AI model downloads in the year to August 2025, narrowly surpassing the U.S.’s 15.86% share, the first time China has done so. According to reports, by early 2026, the President of Congress famous Open-weighted AI models developed in China have increased from approximately 1% of global AI workloads at the end of 2024 to an estimated 30% by the end of 2025.
Much of this momentum is due to an aggressive open source strategy.
Alibaba Cloud flagship Qwen model family surpasses 700 million downloads It will be launched on the developer platform Hugging Face in January 2026, making it the most widely used open source artificial intelligence system in the world. DeepSeek’s R1 inference model, open sourced in January 2025, is said to rival the performance of the best systems in the United States. a fraction of the cost And in the process, it won the goodwill of developers around the world.
Airbnb uses 13 AI models in total, but relies heavily on Qwen, which Chesky describes as “very good,” “fast,” and affordable. Since launching customer service agents, average resolution time has dropped from nearly three hours to six seconds, The company said last year.
It’s also worth noting that the House committee investigating Airbnb’s use of Chinese AI models also sent a similar letter to Anysphere, the company behind the popular Cursor AI coding tool, with investigators focusing on whether Cursor’s Composer 2 can be traced back to Moonshot AI’s line of Kimi models. Additionally, this year’s surge in open source LLM usage driven by Chinese-developed systems include Alibaba’s Qwen series, DeepSeek’s V3 and Moonshot AI’s Kimi K2, according to a report by OpenRouter and venture capital firm Andreessen Horowitz.
Sundar Pichai, CEO of Alphabet Inc., which offers closed-source and open-source models, explain The ultimate question is whether the United States is doing enough to stay ahead.
Pichai said during a fireside chat at the Google I/O developer conference that “the company is optimizing for a lot of factors” and that both open source and closed source models will have a place.
“If it’s open source software with the right license, it doesn’t matter so much where it comes from. With open source software, there’s a community that’s responsible for it and cares about it. So if something goes wrong with the software, it doesn’t go unnoticed. That creates a level of trust for people to adopt the technology.”
“I’m no longer worried about, ‘Do we adopt an open source model from China?’ There’s more question, ‘Are we doing enough in the United States to make sure we stay at the border?'”
It’s a tone that Chesky and other U.S. tech executives appear to share, putting them increasingly at odds with congressional Republicans who appear to view model selection itself as a vulnerability, regardless of where the data flows.
