Six Chinese AI firms accused of aggressively copying US frontier models
Summary
Ars Technica reports that US agencies have accused six Chinese AI firms of industrial-scale distillation of US frontier models, potentially bypassing training costs and expanding capabilities. The piece details attack methods such as bulk account creation to probe model APIs and prompt-injection techniques to reveal internal reasoning, and it outlines recommended mitigations. China denies the claims and the report frames the dispute within broader AI-security and geopolitical competition.