Advertisement
Chinese AI firms defend safety practices, push back on Western criticism
Chinese developers stress cultural context in AI safety, as domestic models narrow the gap with their US rivals to the closest level yet
Reading Time:3 minutes
Why you can trust SCMP
5
Chinese companies are mitigating the risks of artificial intelligence in their own way and should not be judged through a Western lens, according to Chinese industry insiders.
The comments come ahead of what is expected to be a busy month for Chinese AI developers, with new major models set to be released ahead of the Lunar New Year.
Last year, concerns about the risks of Chinese models held back some global users from adopting them, with high-profile DeepSeek in particular banned or restricted in more than 10 countries, including the US, Italy and India.
Advertisement
Chinese AI models entered 2026 having narrowed the performance gap with US rivals to the closest level yet, according to third-party evaluations, prompting calls for Chinese companies to pay greater attention to AI risks such as misuse and misalignment.
In a podcast released on Sunday, former DeepSeek researcher Tu Jinhao said an obsession with catching up with the US had overshadowed domestic work on AI safety.
Advertisement
“All the computational resources are being spent training AI models, with little left to spend on safety work,” said Tu, who was still in high school when he joined the Hangzhou start-up. DeepSeek did not respond to a request for comment.
Advertisement
Select Voice
Select Speed
1.00x