AF: Small: New Directions in Cryptography: Non-Black-Box Techniques against Non-Black-Box Attacks

传统上，加密算法和协议旨在防止通过良好指定的接口（如I/O和通信）与设计算法交互的攻击。然而，目前越来越复杂的计算设备使用方式完全打破了攻击者和被攻击加密算法的“私有内部”之间的传统界限。算法在小型且暴露的机器上运行，这些机器会泄露其内部状态的信息；它们被传送到其他潜在的敌对机器，这些机器可能会检查所有的内部状态，也会错误地报告结果；他们的代码是公开的，并受制于对抗性的修补。该项目旨在开发新的算法和分析技术来处理这一新的现实。这包括能够抵御主机内部状态泄露和篡改的加密算法和协议、程序混淆技术，以及验证在不受信任的机器上完成的计算的技术。这个项目的一个基本前提是，新的分析技术，不再把对手当作黑盒，是必不可少的。因此，开发这些技术需要付出特别的努力。该项目解决了一系列问题，这些问题对现代计算机系统的安全至关重要，因此也对现代社会的福祉和稳定至关重要。但是，即使不考虑实际的适用性，解决的问题是我们对计算概念的理解的核心，代码和数据之间的相互作用，以及算法“理解”任意代码的能力。

Traditionally, cryptographic algorithms and protocols are geared towards protecting against attacks that interact with the designed algorithms via well specified interfaces (such as I/O and communication). However, the increasingly sophisticated ways in which computing devices are currently used completely shatter the traditional boundaries between the attacker and the "private internals" of the cryptographic algorithm under attack. Algorithms are run over small and exposed machines that leak information on their internal state; they are transported to other, potentially adversarial machines which may inspect all the internal state and also misreport the result; their code is exposed and subject to adversarial tinkering.This project is aimed at developing new algorithmic and analytical techniques for dealing with this new reality. This includes cryptographic algorithms and protocols that are resilient to leakage from and tampering with the internal states of the host machines, program obfuscation techniques, and techniques for verifying computation done on untrusted machines. A basic premise of this project is that new analytical techniques, that no longer treat the adversary as black-box, are essential. Consequently, special effort is dedicated to developing such techniques.The project tackles a set of problems that are central to the security of modern computer systems and consequently also to the well-being and stability of modern society. But even disregarding practical applicability, the tackled problems lie at the heart of our understanding of the notion of computation, the interplay between code and data, and the ability to algorithmically "understand" arbitrary code.