Svmuu News: “The White-Haired Stock Guru” Serenity posted that while some of the observations in the UBS report have anecdotal validity, what deserves more attention is the growing number of Chinese reports regarding the distillation of Anthropic’s models. Currently, many U.S. startups and tech companies are leaning toward using cheaper Chinese models (such as DeepSeek) for AI applications, as their per-task costs are significantly lower than those of inference models like Gemini, OpenAI, and Anthropic.
Serenity believes that this trend, driven by capitalism, creates a “classic paradox”—companies will naturally opt for lower-cost solutions, thereby eroding the competitive edge of U.S. models. He proposes that the U.S. must respond by focusing on two areas:
First, build stronger access control and authentication systems, such as “KYC-heavy cutting-edge models” for the U.S. domestic market and tiered access mechanisms for allies, to reduce the risk of model distillation and misuse; At the same time, the U.S. could introduce an identity verification system akin to “bank-level authentication for AI” (such as biometrics combined with short-term access tokens) to raise the threshold for model access and use regulatory measures to restrict account sharing and the resale of access rights.
Second, the U.S. must improve the cost-efficiency of its reasoning models so that they comprehensively outperform competitors like DeepSeek in both price and performance.
Serenity also noted that some high-end models are currently frequently subject to “distillation exploitation”; ideally, additional friction costs should be introduced for accessing models approaching AGI-level capabilities. In summary, the core challenge facing the U.S. AI industry lies in achieving “low-cost inference capabilities” while simultaneously establishing model access security mechanisms comparable to those of the financial system.