Serenity: Anthropic should strengthen model protection instead of blaming policy restrictions on AI competition.

Mars Finance reported on June 28th that, in response to a WSJ report that China's Zhipu AI has caught up with Anthropic in some cybersecurity benchmark tests, market analyst Serenity stated that rather than attributing this result to US government policy, it's more accurate to believe that Anthropic failed to establish a sufficiently robust model protection mechanism to prevent model distillation. Serenity noted that even before Anthropic released Fable, rumors circulated in the industry that Chinese teams were capable of distilling its models. He believes that the key to AI competition lies not only in investing heavily in capital expenditure (CapEx) to build technological moats, but also in protecting core model capabilities and avoiding the "handing over" of achievements through excessive API calls.