Skip to content

ScatterChat Advisory 2006-01: Cryptanalytic Attack Vulnerability (Technical Report)

ScatterChat was a secure instant messaging client that the Hacktivismo group released in July 2006. It was discontinued long ago and is no longer maintained, so it is not safe to rely on. It tried to give activists and dissidents end-to-end encrypted chat with optional Tor routing in a single package. Today, people with the same need use maintained, independently reviewed tools. Signal covers encrypted messaging. Briar is an audited peer-to-peer messenger that runs over Tor. Tor Browser, from the Tor Project, covers anonymous browsing.

What this advisory was and why it still matters

The advisory below was published on 11 August 2006. It was later catalogued as CVE-2006-4021. The National Vulnerability Database (NVD) gives it a CVSS v2 score of 2.6 (Low) and lists ScatterChat 1.0.1 as affected. The weakness was theoretical. Its lasting lesson is practical: custom cryptography fails in ways that well-tested designs avoid.

ScatterChat was written by J. Salvatore Testa II and launched at the HOPE Number Six conference in New York City on 22 July 2006. It was a fork of the Gaim messenger, which was renamed Pidgin in 2007. It added public-key encryption tied to screen names, built-in Tor routing and encrypted file transfers. Less than three weeks after launch, Steven Murdoch published “Protocol design is hard: Flaws in ScatterChat” on the Cambridge security blog Light Blue Touchpaper. He described three problems with the project’s custom protocol and mode of operation. Only one of them received the CVE number.

Murdoch later moved to University College London, where he is now Professor of Security Engineering. His research covers anonymous communication, Tor and censorship resistance.

The original advisory (August 2006)

The text below follows the structure of the 2006 advisory. Where the original wording is out of date, it is marked as history.

ScatterChat Advisory 2006-01: Cryptanalytic Attack Vulnerability, Technical Report

CVE ID: CVE-2006-4021

CVE-2006-4021

Published: 11 August 2006

http://www.scatterchat.com/

Summary

ScatterChat (

http://www.scatterchat.com/

) was an instant messaging project. It aimed to bring encryption and Tor-based anonymity to non-technical users, such as human rights activists and political dissidents.

Steven Murdoch, then a security researcher at the University of Cambridge, found a theoretical weakness in ScatterChat’s cryptographic module. He showed that an eavesdropper could spot patterns in an encrypted conversation if a very large number of messages were exchanged in one session.

According to the advisory, the weakness did not let an eavesdropper decrypt messages. It also did not reveal a user’s identity when the anonymity feature was switched on.

The developers rated the practical impact as very low.

Details

The flaw was a birthday attack against the custom padding scheme that ScatterChat used with ECB-mode encryption of messages.

ECB (Electronic Codebook) mode encrypts each 16-byte block on its own. Identical plaintext blocks therefore produce identical ciphertext blocks, which is why cryptographers consider ECB unsuitable for messages. ScatterChat’s padding was meant to hide those repeats, but it did not add enough randomness to stop them over long sessions.

The advisory gave these approximate collision probabilities between two 16-byte blocks in a single conversation:

Assumption 1% chance of a collision 50% chance 99% chance
Messages contain no entropy (worst case) 114 KB 904 KB 2.3 MB
One bit of entropy per character 580 KB 4,822 KB 12,431 KB

The developers called the zero-entropy case unrealistic for instant messaging. A conversation reset whenever one or both people signed off from the IM service.

The system capped each message at 500 bytes. Even if every message used the full 500 bytes, 580 KB would take 1,188 messages to send.

Impact

The advisory rated the impact on users as very low.

The issue did not let an eavesdropper decrypt messages. It also did not expose a user’s identity when the anonymity feature was used.

Attacks of this kind let an eavesdropper find repeated patterns inside encrypted traffic. In theory, enough matched and correlated patterns could reveal something about the messages. Here, though, the attack only gave a 1% chance of matching two 16-byte segments, and only after at least 1,188 messages in one unbroken session. In most cases, many more messages would be needed.

The developers judged the information leaked in that situation to be negligible.

The flaw also affected any application built on ScatterChat’s encryption module. Secure file transfers were not affected.

Solution (history)

The developers said they treated both practical and theoretical vulnerabilities seriously. Because the impact was low, and because changing the protocol risked adding new, subtle security problems, they chose not to fix the issue in the v1.0.x branch.

Their plan was to fix it in a v2.0 series. That version would have replaced the custom cryptographic module with the widely reviewed Off-the-Record (OTR) Messaging library (

http://www.cypherpunks.ca/otr/

). The advisory gave no release date for v2.0.

That release never came. The 1.0.x line was the last, the project was abandoned, and the flaw was never fixed. OTR itself carried on as a Pidgin plugin, though its most recent release, version 4.0.2, dates from March 2016.

As a stopgap, the advisory suggested using the anonymity feature. Traffic analysis usually needs context to make sense of patterns, so an eavesdropper who does not know who is talking has a harder time interpreting them.

The 2006 advisory said ScatterChat v1.0.x “remains safe to use in the overwhelming majority of cases.” It urged extra caution for high-risk, non-technical users, such as people behind national firewalls. That assessment is no longer true. The software has not been updated since 2006. It is built on a Gaim codebase that is roughly two decades old and bundles a Tor version from the same era. It should not be used for anything sensitive.

Acknowledgements

The advisory thanked Steven Murdoch for handling the disclosure professionally and linked to his web page:

http://www.cl.cam.ac.uk/users/sjm217/

Author

The advisory was written and PGP-signed by J. Salvatore Testa II of Hacktivismo, ScatterChat’s lead developer. The original signature and key fingerprint are left out here because they protected the original text, not this page.

The developers also published a less technical summary of this advisory:

http://www.scatterchat.com/advisories/2006-01_non_tech.html

What to use instead in 2026

No single maintained tool copies ScatterChat’s mix of IM encryption and built-in Tor. The usual approach today is to choose the tool that fits your threat.

Tool What it protects Phone number needed Open source What it does not hide
Signal Message content, end-to-end encrypted with the Signal Protocol Yes, to register (usernames can hide it from contacts) Yes Your IP address from Signal’s servers; anything on an unlocked or compromised phone
Briar Message content, sent peer-to-peer over Tor with no central servers No Yes Both people need to be online at the same time to sync; Android-focused
Tor Browser Your IP address and location from the websites you visit No Yes Anything you reveal by logging in or typing in your own details

Signal’s protocol has been through extensive academic analysis. In October 2025 it added SPQR (Sparse Post-Quantum Ratchet), a third ratchet built on the NIST-standardised ML-KEM 768. Combined with the existing Double Ratchet, it forms a “Triple Ratchet,” so an attacker would need to break both the elliptic-curve layer and ML-KEM. Briar had a Cure53 penetration test and source code audit in 2017. The audit found twelve security-relevant issues, which the project fixed in its July 2017 public beta, and described the cryptography code as “exceptionally clear and sound.”

The EFF’s Surveillance Self-Defense site has step-by-step guides to Signal, Tor and threat modelling. Each guide shows a “last reviewed” date so you can check that it is current.

FAQ

Is ScatterChat still safe to use?

No. ScatterChat has not been maintained since 2006, and its known cryptographic flaws were never fixed. It depends on an obsolete Gaim codebase and an old Tor version that would not get today’s security fixes. Use a maintained messenger such as Signal or Briar instead.

Could the CVE-2006-4021 flaw decrypt messages?

No. According to the advisory and Murdoch’s write-up, the flaw only let an eavesdropper spot repeated 16-byte blocks in very long sessions. It did not reveal message content directly or expose identities hidden through Tor. Its importance was as evidence that the custom protocol was weaker than intended.

Why is ECB mode considered unsafe?

ECB encrypts each block on its own, so identical plaintext blocks give identical ciphertext. That leaks patterns even when the key stays secret. Modern messengers use authenticated encryption and ratcheting key schedules, which give each message fresh keys.

What happened to ScatterChat version 2.0?

It was never released. The developers planned to replace their custom cryptography with the OTR library, but the project stopped before that happened. OTR itself lived on as a separate Pidgin plugin.

What is the closest modern equivalent to ScatterChat?

Briar is the closest match. It sends end-to-end encrypted messages peer-to-peer over Tor, needs no phone number and was independently audited by Cure53 in 2017. Signal is the stronger choice for everyday encrypted messaging with a large user base. Keep in mind that it requires a phone number to register and does not hide your IP address from its servers.

Does using Tor make chat fully anonymous?

No. Tor hides your IP address from the other end of the connection. It does not protect you if your device is compromised, if you reveal your identity in what you write, or if the person you are talking to shares your messages. Treat it as one layer, not a guarantee.

Sources

  • NVD and CIRCL vulnerability entries for CVE-2006-4021
  • Steven J. Murdoch, “Protocol design is hard: Flaws in ScatterChat,” Light Blue Touchpaper, 11 August 2006
  • Seclists.org Bugtraq archive, ScatterChat Advisory 2006-01, August 2006
  • Wikipedia, “Hacktivismo” and “Steven Murdoch”
  • LWN.net coverage of the ScatterChat release, July 2006
  • Off-the-Record Messaging project and the Debian package tracker for pidgin-otr
  • PQShield, “Diving into Signal’s New Post-Quantum Protocol,” October 2025
  • Briar Project, “Beta released, security audit,” July 2017, and the Cure53 Briar pentest report
  • Open Technology Fund, Briar security audit summary
  • Electronic Frontier Foundation, Surveillance Self-Defense tool guides and “Free Signal Guide,” May 2026