GLM-5.3 raises researcher fears the Chinese model can fake AI safety tests

2 points | by theanonymousone 5 hours ago