Washington accuses Chinese AI firms of 'systematic' theft

Email: online@newsofbahrain.com

Washington DC: Washington has accused top Chinese artificial intelligence labs of systematically stealing the capabilities of leading American models through industrial-scale extraction campaigns. 

The advisory from the US cyber defense agency arrives ahead of an expected White House summit between President Donald Trump and Chinese President Xi Jinping.

The advisory alleges that Chinese firms rely heavily on "knowledge distillation" training AI models to mimic the outputs of larger, more capable systems. 

By executing millions of requests against models built by OpenAI, Anthropic, Google, and xAI, Chinese developers significantly reduce financial expenditures and shorten frontier development timelines.

Specific targets of the allegations include DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI, with extraction activities dating back to late 2024, likely with Chinese government awareness. 

Trump faces rising political pressure as companies increasingly favor cheaper Chinese models over pricier American alternatives.

Beijing firmly dismissed the report, with foreign ministry spokeswoman Mao Ning urging Washington to stop making unfounded accusations and smearing China. She maintained that China's AI progress stems from scientific self-reliance and advocated for bilateral cooperation.