注册并分享邀请链接,可获得视频播放与邀请奖励。

Francis
@Francis_Berwa
Builder & investor by modus operandi. Growth @zkPass. Partner @syntaxcapital. A laissez-faire life.
加入 June 2011
898 正在关注    1.4K 粉丝
Anthropic open-sourced FlashProve, a way to prove a response came from a specific model. If agents are going to act for us, that proof matters. You need to know which model answered, and that it stayed inside what it was authorized to do.
显示更多
New Anthropic research: progress on provable inference for large language models. Here, we introduce FlashProve. Building on established cryptographic techniques, FlashProve allows developers to verify whether a given response came from a given AI model. It’s substantially more efficient than prior open-source methods for doing so. These proofs have a range of applications. They can, for instance, make it harder for attackers to hijack a chat, or enable safety evaluators to verify that the model they’re testing is the same one being served to customers. For our full write-up, see here: Further progress in provable inference will likely be an important part of improving trust in AI deployment. To that end, we're open-sourcing FlashProve’s code, here: We’re grateful to our collaborators: Alessandro Chiesa, @GiacomoFenzi, @gwberg95, @miranda__christ, Rahul Dalal ( and Ben Mayer (@transmissions11).
显示更多