Anthropic 发布 FlashProve：用密码学证明 AI 回复来源而不暴露权重
原标题：Anthropic's new prototype proves which model wrote an AI response without exposing its weights
AI 摘要
Anthropic 发布开源研究原型 FlashProve，利用 zkSNARK 密码学技术验证 AI 回复是否确实由声称的模型生成，且不暴露模型权重。该系统在单张 H100 上为 Llama 4 Maverick 生成证明的速度约为每秒 110 个 token，每个证明约 19MB，验证耗时 17 秒，相比此前同类开源系统有数量级提升。Anthropic 认为该技术可用于监管核查、企业安全审计和防篡改，但原型目前仅支持 1024 token 上下文，且只适用于模型选择最可能的下一个 token 的场景，尚不适用于聊天机器人的随机采样输出。
正文节选
Anthropic's new prototype proves which model wrote an AI response without exposing its weights Anthropic has released FlashProve, a research prototype that uses cryptography to verify whether an AI response actually came from the claimed model. The system still has significant limitations. Anyone sending a query to an AI model today has no way to verify that the chosen model actually generated the response. A routing service could quietly swap in a cheaper model to cut costs. An attacker could t