Serenity: Anthropic Should Enhance Model Security Instead of Blaming AI Competition on Policy Restrictions

BlockBeats News, June 28th: In response to a WSJ report claiming that China's Zhìpǔ AI has caught up with Anthropic in some cybersecurity benchmark tests, market observer Serenity stated that instead of attributing this result to U.S. government policy, it should be considered that Anthropic failed to establish an adequate model protection mechanism to prevent model distillation.

Serenity claimed that rumors within the industry about Chinese teams being able to distill Anthropic's model had already circulated before Anthropic released Fable. He believed that the key to AI competition lies not only in massive capital expenditure (CapEx) to build a technological moat but also in protecting the core model capabilities, avoiding handing over achievements through extensive API calls and similar methods.