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

(scmp.com) by theanonymousone | Oct 15, 2026 | 0 comments on HN