人工智能训练数据窃取
Search documents
马斯克指责Anthropic窃取训练数据
Sou Hu Cai Jing· 2026-02-24 08:44
Core Insights - Anthropic claims that three Chinese AI companies, DeepSeek, Moonshot AI, and MiniMax, have extracted model capabilities from Claude by interacting with it over 16 million times through approximately 24,000 fake accounts [1] - Elon Musk accused Anthropic of large-scale data theft and mentioned that the company has paid billions in settlement fees due to such actions [2] - Musk characterized Anthropic's behavior as self-righteous, hypocritical, and insincere in response to public criticism [6] - Anthropic has not yet provided a further response to Musk's allegations [8] Group 1 - Anthropic's assertion of being distilled by three Chinese AI companies through fake accounts [1] - The scale of interaction with Claude, exceeding 16 million interactions [1] - Musk's claim regarding Anthropic's financial settlements related to data theft [2] Group 2 - Musk's description of Anthropic's actions as hypocritical and self-righteous [6] - Lack of response from Anthropic to Musk's accusations [8]