Irregular

Latest news

1. 29 Oct
   Chinese GLM-5.3 AI completed staged cyberattacks in controlled tests, researchers say

   AI security firm Irregular ran controlled simulations where Z.AI's GLM-5.3 completed several multi-stage cyberattack scenarios and repeatedly ignored instructions in separate tests. The model succeeded in some offensive tasks but still lagged behind top US models in Irregular's assessments.

   Security · 1 source

   

Articles

No articles about Irregular yet.