Translated Data: Anthropic has successfully reduced AI discrimination through a new strategy. By incorporating a request in the prompts to treat AI as "extremely friendly," researchers have achieved success in reducing discrimination to nearly zero. This strategy emphasizes the need for cautious use of language models in high-risk decisions, with the influence of decisions to be collectively shaped by the government and society. Additionally, Anthropic is collaborating with Google to deploy Google Cloud TPUv5e chips to support its language model Claude, driving the development of artificial intelligence technology. This research provides a new strategy for reducing AI discrimination, highlighting the careful use of language models in high-risk decision-making.