U.S. agencies say top Chinese AI companies systematically copied American models (www.nbcnews.com)

U.S. intelligence agencies, including the FBI and NSA, have accused leading Chinese AI companies of systematically copying American AI models through a process known as distillation, which has been prevalent since 2024. Major American technologies such as Anthropic's Claude, OpenAI's ChatGPT, Google’s Gemini, and xAI’s Grok are reportedly being mimicked by Chinese firms like DeepSeek, Alibaba, and others. This practice is not only a breach of intellectual property rights but poses a significant threat to the competitive landscape of AI development, as indicated by a detailed report that highlights the targeted distillation of specific functionalities from U.S. models. The implications of this report are critical for the AI and ML community, accentuating the ongoing technological race between the U.S. and China. U.S. officials warn that such large-scale, covert actions undermine American innovation and safety protocols, as echoed by the White House's previous communications. The report offers recommendations for U.S. companies to enhance security and collaboration in response to these challenges, reinforcing the importance of vigilance against intellectual property theft in the rapidly evolving AI sector.