They may have avoided collecting identifying information on specific individual people, but it's absurd to think that they could have reached the scale they did with no knowledge of where and for what purpose their service was being used.

The startup engineer's fantasy is that you can build something useful and users will flock to it through no further involvement of your own, but that's almost never a thing -- and certainly not here.

Anyone operating a crypto mixer knows that people will endeavor to use it to illegally launder funds, and to as I understand it if you know that, you have to implement _effective_ methods to prevent that illegal mixing. It does not matter that there are legal transfers, you have a legal duty to not carry the illegal ones, and if you are not actively trying to prevent those illegal transfers you are liable.

A more day-to-day analogy would be that it is illegal to knowingly buy, sell, or accept stolen property. If you do buy stolen property but there's no reason for you to have thought it is stolen you haven't done anything wrong (though the purchase or sale is invalid, so you've lost the money). If however police can show that you knew or could reasonably know the property was stolen it they could chose to charge you with a crime, and your defense would presumably be some variation of why it was reasonable for you to believe the property was not stolen.

The problem is specifically that everyone running a mixer knows that people attempt to launder their funds through mixers, and so therefore should have mechanisms in place to ensure that they are not accepting transfers that can reasonably related to criminal enterprises, just as you would have to when buying/selling property in the real world. Given that most crypto is trivially auditable there are very few arguments that would support not having at minimum automated rejection of transactions involving any known wallet, and moreover constantly updating the list of known bad addresses.

Certainly if I were on a jury, and the person running the mixer could not show their reasonable actions to prevent laundering, that the system was constantly tracking, updating, and expanding the disallowed wallets precisely because that can be done almost entirely automatically, that would count against their claims to not be intentionally supporting laundering.

Again, the issue here is not running the mixer, it's not that occasional illegal transactions go through - that's unavoidable - it's the failure to take reasonable steps to stop plausibly illegal transfers. The way banks handle this is they block suspicious transactions pending receipt of the actual sender and recipient, and documentation of the origins of the funds. Which is what a mixer would need to do.

If it was just a message sent by an agent, I hope we will not see a conviction.

Or rather, even if the admins were knowingly facilitating crime, from a liability perspective they would be better off reporting/rejecting any users that say they're committing a crime and only do business with criminals smart enough to wink wink nudge nudge.

I can only think of two scenarios where someone would tell a mixer the money is dirty:

1. They are law enforcement

2. They have already been caught by law enforcement and are being used as bait, either knowingly or not

The issue is that if you run a mixer and someone says "please launder my drug money" you reject that person's coins unless you like Federal prison. It is like if you run a gun store and someone comes in and asks for a gun "to rob a bank", that person is a Federal agent and if you sell them a gun an indictment is forthcoming.

People are doing life sentences for giving their friend a ride when their "friend" decided to hold up a liquor store.