China newsletter icon
From Semafor China
In your inbox, 1x/Week
Sign up

Chinese open-weight models reignite AI safety debate

Jul 28, 2026, 11:27am EDT
PostEmailWhatsapp
KIMI illustration
Florence Lo/Reuters

The rapid rise of Chinese open-weight AI reignited the US debate over how to balance safety and security risks while driving the growth of the American tech ecosystem.

Some fear that Silicon Valley will get hooked on cheap Chinese AI that’s accused of ripping off American labs, introducing “various baskets of risk,” an American Enterprise Institute scholar wrote this year.

The US is now threatening sanctions against Chinese AI firms and Beijing is vowing to retaliate. But the safety question isn’t about who makes an AI model, it’s about “who owns and operates” the server it runs on, China tech expert Samm Sacks argued, potentially reassuring American firms and developers who are “worried about data being sent to China.”

AD