Leading Democrat asks AI firms for data on any Chinese access to sensitive code
XLK•Ro Khanna asked OpenAI, Anthropic, Google, Meta and SpaceXAI for data on known efforts by China or other hostile actors to access their model weights illegally, and information on cybersecurity measures to prevent theft.
1. Khanna seeks information
U.S. Representative Ro Khanna asked five AI companies to provide data on known efforts by China or other hostile actors to illegally access their model weights, as well as information on their cybersecurity measures. In letters to the companies’ CEOs, he warned that theft by China could erode America’s AI lead.
2. Distillation concerns
OpenAI and Anthropic have reported multiple instances of Chinese AI firms training their own tools using outputs from leading Western models, a practice known as distillation. The article says there are few known cases of bad actors stealing model weights, the parameters and code that make an AI model operate.
3. National security debate
Khanna said the national security and economic prospects of the United States depend on the cybersecurity of a small number of companies. The letters add to concerns among lawmakers about AI risks; President Donald Trump has downplayed some concerns and said tech executives agreed to establish voluntary AI standards.
