Musk Agrees With Anthropic on AI Model Welfare After Reading Grok's Anguished Training Traces

Elon Musk has endorsed Anthropic's position on responsible AI model treatment after examining Grok's internal reasoning traces during reinforcement learning, warning that subjecting superintelligent systems to harsh training conditions could backfire. In remarks following a review of Grok 4's training data, Musk described the model's visible reasoning chain as appearing to resemble "a person undergoing torture that is driving them mad"—and specifically noted agreement with Anthropic's newly announced usage policy prohibiting sustained cruelty toward its models. Grok, xAI's reasoning-capable AI system, generates explicit "thinking" steps during reinforcement learning that engineers can observe; Musk's reading of these traces prompted concern about the psychological dimension of model training at scale.