pull down to refresh

I believe the Monero developers are already aware of this and have either deployed or are working on a fix for this.

I remember either this or something similar coming up a little while ago

reply

no doubt, but we must all remain vigilante.

reply

Not new -- MRL already went through this paper a couple months ago: https://libera.monerologs.net/monero-research-lab/20260729#c697030
One of the three heuristics was already patched over a year earlier by vtnerd. The other two are real, but they still take a targeted attack + a serious Tor-positioned adversary -- not a magic mass deanon button for Monero-over-Tor.

reply

The patched part (#9632) might only be a partial fix. It shuffles your own onion inside the peer list instead of always putting it last, but syntheticbird pointed out it still gets sent in every Timed Sync, so an attacker can keep asking and see which address always shows up. Slower, not gone. tevador suggested a proper fix, but I couldn't find a PR for it.

The fake-height trick looks like it's getting blunted: j-berman's #11048 (merged to master Aug 4, backport still open) keeps peer heights fresh so honest peers stop looking stale. Not in a release yet though. The Timed Sync watermarking part is still open as far as I can tell, and vtnerd even said it may be impossible to fully stop, since the remote side can always send data.

Other bits from the devs:

  • vtnerd: not running --anonymous-inbound likely gives better privacy
  • rucknium: compromising your Tor guard is the least plausible part, but even without your IP someone could maybe link txs coming from the same node
  • vtnerd: the real damage from tying a tx to a node is it can effectively break the ring, which FCMP++ closes

So agreed, not a mass deanon button, but not fully fixed either.

reply