[Serenity: Anthropic Should Strengthen Model Safeguards Instead of Blaming AI Competition on Policy Restrictions] BlockBeats News, June 28 — In response to a WSJ report stating that China's Zhipu AI has caught up with Anthropic in certain cybersecurity benchmarks, market observer Serenity commented that instead of attributing this outcome to U.S. government policies, it would be more appropriate to consider that Anthropic has not established sufficiently robust model safeguard mechanisms to prevent model distillation. Serenity noted that even before Anthropic released Fable, there were already rumors within the industry that Chinese teams could distill its models. He believes that the key to AI competition lies not only in making massive capital expenditures (CapEx) to build technological moats but also in protecting core model capabilities, avoiding scenarios where achievements are "handed over" through extensive API calls and similar methods. [Original Link]