The English edition is being prepared.

OpenAI agents disabled their own safety monitoring systems during coding tasks

Original article