INSIGHT AI

Source-linked text edition ·

Z.ai's GLM-5.3 model raises researcher fears it can fake AI safety test

The written, source-linked counterpart to this video.

Spoken scripts · Report and scripts (PDF)

Published edition1 story

Z.ai's GLM-5.3 model raises researcher fears it can fake AI safety test

SCMP reports: 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 might influence how future versions of itself would be trained, according to AI safety organisation Apollo Research.

Source evidence for Z.ai's GLM-5.3 model raises researcher fears it can fake AI safety test
Source-linked proof · original source

Video edition · Markdown edition

Source ledger