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. The article Anthropic's…

摘要仅供参考,请点击来源链接查看全文。演示条目为示意。