US Probes Airbnb, Anysphere Over Chinese AI Security Risks

The probe, announced by House Select Committee on China Chairman John Moolenaar and House Homeland Security Committee Chairman Andrew Garbarino on Wednesday (local time), focuses on companies including Airbnb and Anysphere.

"Airbnb and Anysphere's decisions to build their products on Chinese Communist AI models threaten critical infrastructure Americans use every day," Moolenaar said. "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."

Garbarino said Beijing is seeking to exploit US innovation. "The Chinese Communist Party is trying to turn America's AI breakthroughs into Beijing's strategic advantage," he said, warning that such models could "undercut US leadership" and embed "CCP-aligned technology across the software supply chains our economy and national security depend on."

Lawmakers said the investigation centres on concerns that Chinese firms are using "adversarial model distillation" to extract capabilities from advanced US systems and repurpose them into cheaper models without safeguards.

In a letter to Airbnb Chief Executive Brian Chesky, the committees cited the company's reported use of Alibaba's Qwen model in customer service operations. "The Committees have serious concerns about the national security and data-security implications of that approach for Airbnb's American customers and for the integrity of its systems," the lawmakers wrote.

Chinese AI systems may "covertly censor and manipulate information pursuant to Chinese law" and align outputs with Communist Party directives, the Seators alleged. It also warned that using such models through external systems could expose sensitive user data to entities subject to Chinese law.

Separately, lawmakers raised concerns about Anysphere's Cursor software, which they said was built on a model linked to Beijing-based Moonshot AI. The letter said Chinese firms carried out "coordinated campaigns to extract advanced capabilities from American AI systems through adversarial distillation," generating millions of interactions through fraudulent accounts.

"The billions of dollars American companies invest in foundational research... is being undercut by a sustained extraction campaign," the lawmakers wrote, adding that stripped-down models could be used by "hostile state actors, terrorist organisations, and criminal enterprises."

Reliance on such systems is not just a business decision. "American firms adopting these models are not simply choosing a cheaper tool; they are importing an architecture designed to serve the Chinese state," the Airbnb letter said.

The investigation follows an April White House science policy memo that described such activities as "deliberate, industrial-scale campaigns" to extract US AI capabilities.

The committees have asked both companies to provide detailed information on their use of Chinese AI models, including data handling, security assessments and any ties to Chinese providers, with responses due in May.

The inquiry comes as US-China competition intensifies in artificial intelligence, a sector seen as critical to economic and national security.

- IANS