MaxRa comments on My AGI Threat Model: Misaligned Model-Based RL Agent