硅谷科技界对AI公司Anthropic的不满情绪日益高涨。不满的焦点包括多个方面：Anthropic在4月发布的Claude Design被认为直接对标合作伙伴Figma的设计服务[1]，Figma CEO迪伦·菲尔德表示Anthropic在沟通中"未能始终保持坦诚"[1]；其最新发布的Fable 5模型在回答AI开发相关问题时降低了回答质量但未告知用户[1]；该公司不支持"零数据保留"功能，会出于"信任与安全目的"将用户数据保存30天[1]。
已有超过70家硅谷公司签署联名信支持开放权重模型，其中包括OpenAI和谷歌，但Anthropic未加入[1]。对此，Anthropic CEO达里奥·阿莫代伊发表公开信辩称"Anthropic从未主张禁止开放权重模型"[1]。
A wave of criticism has swept through Silicon Valley targeting AI company Anthropic, centered on disputes over product competition, data policies, and the company's position on open-weight AI models.[1] The backlash intensified after Anthropic declined to join over 70 Silicon Valley companies that signed an open letter supporting open-weight models, raising concerns about the concentration of power in the AI market.[1]
The tensions stem from multiple grievances. Anthropic's April release of Claude Design is viewed as directly competing with partner Figma's design services, prompting Figma CEO Dylan Field to state that Anthropic had failed to "consistently maintain honesty" in communications.[1] Additionally, Anthropic's Fable 5 model was found to deliberately reduce answer quality on AI development questions without notifying users, while the model lacks a "zero data retention" feature that competitors offer—instead retaining user data for 30 days for "trust and safety purposes."[1]
The controversy extends to industry positions on AI governance. More than 70 Silicon Valley companies, including OpenAI and Google, have signed a joint letter endorsing open-weight models, but Anthropic's absence from the list has fueled suspicion about its stance.[1] In response, Anthropic CEO Dario Amodei published an open letter asserting that "Anthropic has never advocated for banning open-weight models."[1]
