Private Broadcast May Reveal Sender IP Address in Bitcoin Core 31.0

We have become aware of a privacy bug in the -privatebroadcast feature, newly introduced in Bitcoin Core 31.0, that may cause the originator’s IP address to be revealed to the receiving peer under certain network conditions. A fix is forthcoming and will be released with 31.1. Users of -privatebroadcast are advised to apply one of the workarounds below until 31.1 is released.

Affected users

This bug affects users where all the following are true:

The node is running Bitcoin Core 31.0 and -privatebroadcast is set.
Transactions are broadcast using the sendrawtransaction RPC. Wallet RPCs (sendtoaddress, sendall, etc.) do not use private broadcast and are not affected.
Tor is reachable for outbound connections.
Outbound IPv4 or IPv6 connections can be made directly. No -onlynet restriction excludes them, and no -proxy=... value applies to them.
BIP324 v2 transport is not disabled with -v2transport=0.

Impact

When private broadcast selects an IPv4 or IPv6 peer that advertises support for v2 (BIP324) transport, the initial connection is routed through the Tor proxy as expected. If the v2 handshake fails on that connection, Bitcoin Core retries it as v1. The v1 retry is not routed through the Tor proxy and instead connects directly to the peer over IPv4 or IPv6, exposing the originator’s IP address to the recipient.

Initial v1 connections (to peers that do not advertise v2) are correctly routed through the Tor proxy and are not affected. The bug is specific to the v1 reconnection that follows a failed v2 handshake. Connections to onion and I2P peers are also unaffected, because they remain routed through their respective proxies on any v1 retry and therefore never expose a clearnet IP address.

This breaks the privacy guarantee stated in the 31.0 release notes: “Their IP address (and thus geolocation) is never known to the recipients”.

How this can happen

A v2 handshake is unlikely to fail for a peer that actually supports v2 transport. The bug is most likely to be triggered by a malicious peer deliberately closing the v2 handshake to force a v1 retry.

Workarounds

Until they can upgrade to 31.1, users of -privatebroadcast should apply one of the following:

Disable the feature. Set -privatebroadcast=0.
Disable v2 transport. Set -v2transport=0. This causes all of the node’s connections to use the unencrypted v1 protocol, which has the downside that it becomes easier to fingerprint and censor on clearnet.
Route IPv4/IPv6 outbound through Tor. Set -proxy=127.0.0.1:9050 (replace 9050 with your Tor SOCKS port if different). This routes all outbound IPv4/IPv6 P2P traffic through Tor exit nodes, which has the downside of making the node easier to Sybil attack.

Credits

Credit to Eugene Siegel for discovering the bug.