Model substitution + false token reporting, users still think they are using GPT-4.

This article analyzes the rise and risks of AI transfer stations, revealing how they profit through methods such as bulk registration to obtain free quotas, refund arbitrage, false reporting of tokens, model swapping, and data resale. While their prices are low, they pose risks of illegal operations, lack of data security obligations, and infringement on personal information. The author emphasizes that the industry should move toward standardization, protect user data, clarify authorization, and eliminate data trading to promote vendors and practitioners to maintain the industry's long-term healthy development in a compliant and responsible manner.