It's very good. It's also fast and cheap.
In an era when the architecture of intelligence itself has become geopolitical terrain, two US House committees have turned their scrutiny toward Airbnb and Anysphere, questioning whether the pursuit of cheaper, faster AI has quietly opened American user data to Chinese influence. The investigation reflects a deeper tension running through the modern technology economy: the logic of the market and the logic of national security do not always point in the same direction. What began as a CEO's candid admission about cost and performance has become a flashpoint in the broader reckoning over how deeply Chinese technology has woven itself into the fabric of American digital life.
- Republican-led House committees have launched a formal joint investigation into Airbnb and Anysphere, demanding answers about why these American companies chose Chinese AI models over domestic alternatives.
- Airbnb CEO Brian Chesky's own public praise of Alibaba's Qwen model — chosen for its speed and low cost over OpenAI's ChatGPT — has become the evidentiary centerpiece of congressional concern.
- Lawmakers warn that Chinese AI models carry hidden vulnerabilities and expose American user data to Beijing's censorship apparatus, framing cost-driven tech decisions as potential national security liabilities.
- The open-source nature of Chinese AI tools is itself under suspicion, with committees arguing that their affordability comes at the price of bypassing the safety and regulatory standards American companies must uphold.
- The inquiry is escalating: employees from both companies may be called before Congress in person, signaling that written inquiries alone will not satisfy lawmakers seeking accountability.
Two Republican-led House committees — Homeland Security and the China Select Committee — have jointly launched an investigation into Airbnb and Anysphere, the maker of the AI coding tool Cursor, over their use of Chinese artificial intelligence systems. Formal letters have been sent to both companies requesting detailed explanations of their choices, their reasoning, and any communications with Chinese AI developers.
At the heart of the inquiry is a decision Airbnb CEO Brian Chesky openly defended just months ago. In December 2025, Chesky acknowledged that Airbnb had built its customer service agent on Alibaba's Qwen model rather than OpenAI's ChatGPT, citing Qwen's superior speed and lower cost. Despite his personal friendship with OpenAI CEO Sam Altman, Chesky indicated no urgency to deepen the ChatGPT partnership, noting that seamless integration would require capabilities OpenAI had yet to develop.
Lawmakers are now framing that pragmatic business decision as a potential national security risk. Representative John Moolenaar stated plainly that Chinese AI models are shaped by Beijing's censorship regime and introduce hidden vulnerabilities that endanger American data. Committees have also raised concerns that the cost advantages of open-source Chinese AI tools come precisely because they sidestep the regulatory and safety standards binding American competitors.
The investigation may not stop at letters. Reports indicate that employees from both Airbnb and Anysphere could be summoned to congressional briefings in person — a notable escalation at a moment when US-China tensions are actively reshaping technology policy. Meanwhile, Airbnb continues expanding its platform, having recently introduced social features allowing guests to connect after shared experiences, even as questions deepen about the AI infrastructure powering its customer-facing systems.
Two Republican-led House committees have launched a joint investigation into Airbnb and Anysphere, the company behind the AI coding platform Cursor, over their reliance on Chinese artificial intelligence models. The House Homeland Security Committee and House China Select Committee, chaired by John Moolenaar of Michigan and Andrew Garbarino of New York respectively, have sent formal letters to both companies requesting detailed information about their use of Chinese AI systems, the reasoning behind those choices, and any communications they've had with Chinese AI developers.
The investigation centers on a decision Airbnb made that its CEO Brian Chesky publicly defended just months earlier. In December 2025, Chesky explained that Airbnb had chosen to power its customer service agent using Alibaba's Qwen model rather than integrating with OpenAI's ChatGPT. His reasoning was straightforward: Qwen was faster and cheaper. "We're relying a lot on Alibaba's Qwen model. It's very good. It's also fast and cheap," Chesky said at the time. He acknowledged that while Airbnb does use OpenAI's latest models, they tend to rely more heavily on the Chinese alternative because of the cost and speed advantages it offers over American competitors.
The timing of Chesky's comments is notable given his personal relationship with Sam Altman, OpenAI's CEO. Despite their friendship, Chesky indicated that Airbnb would not be rushing into a deeper partnership with ChatGPT. He suggested that OpenAI would need to develop more robust integration capabilities before Airbnb would consider embedding ChatGPT directly into its platform in a way that felt seamless to users. He also noted that Airbnb's community-based model, with its verified member system, would require any AI integration to operate in an almost self-contained manner within the chatbot environment.
Now, lawmakers are raising alarms about the national security implications of that choice. In their letters, the committees have expressed concern that sharing large volumes of user data and business information with Chinese AI companies poses risks to American interests. Representative Moolenaar was direct about the worry: "The AI models these companies use are trained by China's censorship regime and introduce hidden vulnerabilities that put Americans' data and businesses at risk." The committees are particularly concerned about the open-source nature of many Chinese AI tools, which lawmakers argue offer cost savings precisely because they bypass the regulatory and safety standards that American companies must meet.
The investigation may extend beyond written inquiries. According to reports, employees from both Airbnb and Anysphere could be summoned to attend in-person congressional briefings to explain the technical and business decisions that led to their adoption of Chinese AI systems. This represents an escalation in congressional scrutiny of American tech companies' foreign AI dependencies at a moment when geopolitical tensions between the United States and China are shaping technology policy.
Airbnb has been expanding its AI capabilities in other ways as well. In December 2025, the company rolled out a new feature allowing guests to share their Airbnb profiles with other travelers after booking an experience, and users who have taken the same tours can now message one another directly. The company implemented privacy protections so that conversations only continue if the recipient accepts the initial message request. These moves suggest Airbnb is actively building out its platform's social and communication features, even as questions mount about the infrastructure powering its customer-facing AI systems.
Citas Notables
The AI models these companies use are trained by China's censorship regime and introduce hidden vulnerabilities that put Americans' data and businesses at risk.— Representative John Moolenaar, House Homeland Security Committee chair
La Conversación del Hearth Otra perspectiva de la historia
Why would Airbnb choose a Chinese AI model when American alternatives exist? The cost difference must be significant.
It's not just cost—it's the whole package. Qwen is genuinely faster at inference, meaning it responds quicker. For a customer service agent handling thousands of queries, speed matters as much as price. OpenAI's models are more powerful but also heavier. Chesky was being honest about the trade-off.
But Chesky is friends with Sam Altman. Wouldn't that relationship push him toward ChatGPT?
Friendship and business are different things. Chesky said OpenAI's integration tools weren't ready yet. He wasn't rejecting Altman—he was saying the product didn't fit Airbnb's needs at that moment. The real question is whether he should have waited.
What's the actual security risk here? Is it that China could spy on Airbnb users?
The committees are worried about two things: one, that user data flows through Chinese servers or systems; two, that the model itself might have vulnerabilities built in by design. They're also concerned about the principle—that American companies shouldn't be dependent on Chinese infrastructure for critical functions.
Could Airbnb have built its own AI system instead?
Possibly, but that's expensive and slow. Building a competitive large language model takes years and billions. Buying or licensing an existing one is faster. The question Congress is asking is whether speed and cost should override national security concerns.
What happens if Airbnb's employees are called to testify?
They'll have to explain the decision-making process in detail—who evaluated which models, what the cost comparisons were, whether anyone raised security flags internally. It becomes a record. And it signals to other companies that this choice might not be safe to make.