The Decoder· Manuel Uth·· 4 天前AI 评分66
Anthropic 称新防御将 Claude 推理 token 窃取量削减最高 99%
Anthropic says it cut Claude reasoning token theft by up to 99% in fight against industrial-scale distillation
AI 导读
Anthropic 称其新防御措施将 Claude 推理 token(模型内部思维链步骤)被提取的量削减了 90% 到 99%,成功蒸馏尝试在其防御上线后大幅下降。
来源:The Decoder · the-decoder.com