Anthropic开源FlashProve实现大模型推理可验证
New Anthropic research: progress on provable inference for large language models…
大模型安全基础设施的重要进展，提供可复现的验证工具，值得安全与工程团队关注。
New Anthropic research: progress on provable inference for large language models.
Anthropic 最新研究：大型语言模型可证明推理的进展。
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.
在此，我们介绍 FlashProve。FlashProve 建立在成熟的密码学技术之上，允许开发者验证给定响应是否来自特定的 AI 模型。其效率远高于此前用于此目的的开源方法。
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.
这些证明具有多种应用场景。例如，它们可以增加攻击者劫持对话的难度，或使安全评估人员能够验证他们正在测试的模型与提供给客户的模型是同一个。
For our full write-up, see here: https://www.anthropic.com/research/provable-inference
完整文章请参见此处：https://www.anthropic.com/research/provable-inference
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: https://github.com/anthropics/FlashProve
在可证明推理方面的进一步进展，很可能将成为提升对 AI 部署信任度的重要组成部分。为此，我们开源了 FlashProve 的代码，地址如下：https://github.com/anthropics/FlashProve
We’re grateful to our collaborators: Alessandro Chiesa, @GiacomoFenzi, @gwberg95, @miranda__christ, Rahul Dalal (http://rdalal.github.io) and Ben Mayer (@transmissions11).
我们要感谢我们的合作者：Alessandro Chiesa、@GiacomoFenzi、@gwberg95、@miranda__christ、Rahul Dalal（http://rdalal.github.io）和 Ben Mayer（@transmissions11）。
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力