US Officials Allege Chinese AI Firms Secretly Copied US Models

US government agencies have accused Chinese AI companies of covertly extracting billions of tokens of data from major US artificial intelligence platforms, including OpenAI, Anthropic, Google Gemini, and Grok. The technique, known as model distillation, allegedly allowed these firms to build competing AI systems more cheaply by learning from the outputs of established models rather than developing their own from scratch.

While this issue centres on intellectual property and international AI competition rather than a direct cyberattack on businesses, it highlights a broader trend: valuable data and AI outputs are increasingly targeted by sophisticated actors seeking to shortcut development costs. For small and medium businesses using AI tools or exposing APIs, this is a reminder that data access controls and monitoring of unusual usage patterns matter, whether the threat comes from criminals or state-linked actors.