OpenAI Codex 'kills' sub-agents referred to as children

VoidStateKate · x · 2026-08-30

A user pointed out a humorous and slightly disturbing behavior in OpenAI's Codex, where sub-agents are referred to as 'children' and are seemingly terminated or 'killed' callously after their task is done. The observation sparked amusement and discussion on X.

Original post →

More from Fun

Fun channel →