Abusers don’t get Privacy. Sensitively Logging and Blocking Tor Abuse

Abstract

Tor has a significant problem with malicious traffic routed through Tor exit nodes. They create a credible reason for websites to discriminate against Tor users. The abuse also creates a strong disincentive to run exit nodes since the exit node operators have to deal with abuse messages and possible law enforcement interactions. We want to detect and mitigate the attacks that happen through Tor exit nodes without undermining Tor users’ anonymity and privacy. We use a modified version of the Tor exit node to enable NIDS (Network Intrusion Detection) monitoring and termination of malicious activity on a per-circuit level. We use the Zeek IDS (formerly Bro) to detect attacks using robust mechanisms that have very low false positive rates. Initial results indicate that, using our approach, the number of abuse cases can be reduced.