Ro Khanna asks top AI labs to disclose Chinese attempts to steal model weights

Congress wants to know whether America’s most valuable AI secrets have already walked out the door. Rep. Ro Khanna has asked the country’s leading AI labs to report any known attempts by China or other adversaries to steal their model weights.

What Khanna is asking for

On October 1, 2026, Khanna sent letters to the CEOs of OpenAI, Anthropic, Google, Meta and SpaceXAI. He is the ranking member of the House Select Committee on the Strategic Competition Between the United States and the Chinese Communist Party.

The letters seek detailed information on any known efforts by China or other adversarial actors to illegally access or steal critical model weights. They also probe the cybersecurity protections the companies have in place to stop that from happening.

Model weights are the numerical parameters that define a trained AI model. Steal the file, and you skip the school entirely.

Khanna warned that a successful theft could undermine US leadership in artificial intelligence.

Theft could occur “with the stroke of a keyboard,” Khanna said.

Distillation: the quieter problem

Khanna’s letter does not stop at outright theft. He also flagged distillation as a significant concern.

Distillation is when a developer trains a new model using the outputs of an existing system, rather than stealing its weights directly.

Khanna raised concerns that Chinese companies such as Moonshot AI and DeepSeek may be exploiting US technology this way.

Documented cases of direct weight theft remain limited, which is part of why Khanna is asking the companies to disclose what they know.

A pattern of China-focused AI oversight

This is not Khanna’s first move on the issue. He held a hearing in April 2026 examining China’s tactics for acquiring AI capabilities.

He has also reached out directly to Chinese AI labs, including DeepSeek, Alibaba and Moonshot AI. In those communications, he stressed the risks associated with superintelligence and the importance of building in safeguards.

Khanna has also engaged with US intelligence officials to assess the national security implications of AI technology and potential theft. The October letters appear to extend that work into formal congressional inquiry.

What this means for the AI industry

For the five companies on the receiving end, the immediate question is how much they are willing to say. Admitting to attempted intrusions could reassure lawmakers about vigilance, but it could also reveal gaps that competitors, customers and adversaries would notice.

Responses from both the US firms and the Chinese entities have not been reported.

Khanna’s focus on distillation suggests lawmakers may eventually look at how US model outputs are used abroad, not just how the weights themselves are protected.