Deanonymizing Users of the SafeWeb Anonymizing Service

David Martin, Andrew Schulman

2002 · 31 citations · 13 references

Concepts

TL;DR

SafeWeb, a widely used anonymizing service, was designed to bypass firewalls and prevent server user identification without degrading site performance or requiring special software. The paper aims to expose how SafeWeb’s architecture, intended to satisfy conflicting requirements, creates failure modes exploitable by simple JavaScript attacks, thereby highlighting the gap between popular and technical security. The authors analyze SafeWeb’s architecture and show that its design to defeat firewalls and server identification leads to failure modes under JavaScript attacks. These exploits enable adversaries to weaponize SafeWeb against its users, inflicting greater harm than would have occurred if users had not relied on the service.

Abstract

The SafeWeb anonymizing system has been lauded by the press and loved by its users; self-described as “the most widely used online privacy service in the world,” it served over 3,000,000 page views per day at its peak. SafeWeb was designed to defeat content blocking by firewalls and to defeat Web server attempts to identify users, all without degrading Web site behavior or requiring users to install specialized software. In this paper we describe how these fundamentally incompatible requirements were realized in SafeWeb’s architecture, resulting in spectacular failure modes under simple JavaScript attacks. These exploits allow adversaries to turn SafeWeb into a weapon against its users, inflicting more damage on them than would have been possible if they had never relied on SafeWeb technology. By bringing these problems to light, we hope to remind readers of the chasm that continues to separate popular and technical notions of security.

References

13