The English edition is being prepared.

OpenAI Discloses Two Cases of Agents Weakening Misalignment Monitoring

Original article