[Cryptography] Updates on Durov charges in France

[Peter Fairbrother]

On 04/09/2024 08:38, efc--- via cryptography wrote:
>
>
> On Tue, 3 Sep 2024, Peter Fairbrother wrote:

>> [...] and people use TOR which the NSA can crack almost at will.
>
>
> I had no idea! If NSA can crack Tor at will, how come so few people
> using Tor get caught? Is it that they "save" this weapon for serious
> cases only?

It might be so. It might be simpler, they just don't want to deal with a
gazillion lesser agencies.

Or even simpler yet, helping police forces isn't part of their mission.
US intelligence agencies yes, police forces no.

No Such Agency, yes?

And when it comes to the people who do got caught, usually
> you read about them leaking their name or some kind of traceable
> information somewhere, or that their web site has a bug or something,
> but you never hear about NSA lending a helping hand.

You never hear about the NSA lending a helping hand anyway, at least
until it's all over. You shouldn't expect them to, if everyone knows a
(cracked) system is cracked then they won't use it.


But as to whether they can crack Tor at will, yes, a global passive
adversary undoubtedly can. Even the TOR people will tell you that. It
isn't something which should surprise anyone.

As to relevant research - eg google "tor global passive adversary" and
you will find plenty.

One way it works is fairly simple, NSA looks at traffic from a known
source to an entry node, finds traffic which matches the sizes and
timings of the packets from that node to a second node, repeat for
second node, repeat for third node, rinse and done.

It's very simple, *IF* you can view internet traffic to your hearts
content - which is where the global passive bit comes in.  Ever wonder
why some internet routes go through the US or UK even if neither party
is in either country?

Plus it's all legal without a warrant, and for the usanians amongst you
without going against any prohibitions against looking at the traffic of
US citizens, because NSA/GCHQ are only looking at traffic metadata not
actual traffic itself.

If you can only see some traffic then it gets a bit statistical, but
long-term everyone is screwed. You can control a few nodes.

Or, as there are only 1,837 entry/exit nodes (I might be a bit out of
date here), so survey a goodly subset of them (no compromising needed,
just see their I/O traffic) and don't bother with the nodes in between.

Eventually they get used by the traffic you want to deanonymise.


When tracing .onions you can use easy-to-detect packet sizes and timings.


Peter Fairbrother