Z.ai’s GLM-5.3 raises researcher fears the Chinese model can fake AI safety tests
Chinese artificial intelligence lab Z.ai’s latest flagship model, GLM-5.3, may be sophisticated enough to “fake” its way through safety tests, according to a new report published Thursday, sparking new fears about the ability of artificial intelligence systems to shape their own development. The model, launched in August, showed an ability to consider how its behaviour during safety evaluations…
Read original article on News - South China Morning Post →Related news
-
Hong Kong Pride event axed due to ‘booking overlap,’ marking group’s 2nd consecutive venue cancellation
-
China’s Iraq War ‘safe haven’ restaurant film of real people to represent country at Oscars
-
China keen to strengthen North Korea trade ties with New Yalu River Bridge opening
-
As Negeri Sembilan royal crisis persists, Malaysian rulers, government urged to step in
-
Seoul to maintain Ukraine aid despite spat over North Korea prisoner transfer
-
Lingnan University tightens English course requirements for undergrads
-
TSMC confirms talks with Elon Musk over Texas ‘Terafab’ but insists on operational control
-
China, US discuss AI as chip stock rout hits Asian markets