China says US accusations of AI theft are ‘groundless’

On Wednesday, China dismissed US allegations that Chinese AI labs were stealing American models' capabilities as "groundless", amid a tech rivalry. The accusations were made by the US Cyber Defense Agency in an advisory released on Tuesday. It targeted prominent Chinese AI firms like DeepSeek and Moonshot AI, alleging they systematically stole knowledge via large-scale imitation campaigns.

This news came before a scheduled meeting between President Donald Trump and Chinese President Xi Jinping at the White House. China labeled the US accusations as "groundless", accusing Washington of using the crackdown on knowledge distillation to gain an industrial monopoly. Distillation, a method of training AI models to replicate larger ones, is common, but the US claims Chinese firms have scaled it to copy American systems illicitly.

The US advisory also detailed how firms like DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI allegedly used millions of interactions to steal US models from OpenAI, Anthropic, Google, and xAI. Chinese entities employing this strategy benefit from quicker AI development and lower training costs, according to the advisory.

These actions allegedly occurred since late 2024, with government awareness. The Chinese companies contacted by AFP, except StepFun, did not respond. Moonshot AI declined to comment. The US President is under pressure to respond to Chinese AI models' growing popularity and cost-effectiveness compared to US counterparts. In July, Chinese startups wrote to the Trump administration urging against outright bans on foreign AI models, recommending restrictions only for government use.

Beijing retaliated by accusing US AI systems of using Chinese examples to train their models. At a Wednesday briefing, Chinese Foreign Ministry spokeswoman Mao Ning urged the US to stop making "unfounded accusations" against China. She highlighted China's self-reliance in AI development and the strength it brings to both nations' scientific and technological capabilities. Both countries, she said, are major players in AI and should enhance cooperation.