Interview with Noah Perlman, Chief Compliance Officer of Binance. We explore the future of crypto compliance and the evolving role of humans as artificial intelligence becomes more deeply integrated into detection and monitoring systems. 🔹 Why false positives in compliance are expected to decrease significantly 🔹 How AI models are trained to recognize patterns with high levels of certainty 🔹 Which alerts could be resolved without human intervention while still meeting regulatory standards 🔹 Why validation remains critical before fully autonomous AI decision-making 🔹 How human experts will shift toward investigating more complex and sophisticated threats 🔹 Why compliance remains a constant “cat and mouse” game despite advances in AI As AI takes over more routine processes, the role of humans in crypto compliance is not disappearing — it is evolving toward higher-value, more strategic work.
Full Transcript
Obviously, you mentioned we'll have a reduction in the number of false positives, which obviously is a massive headache. I don't know, I mean it's a We I think there's a lot of data in the compliance world. I don't know in the crypto compliance world how much money is wasted every year just cost just dealing with these false positives. How do you think we'll be dealing with this? You think it'll be mainly done via AI will be able to enable us to reduce the number of false positives? Uh and and especially what is the role of humans that you see in the next let's say 5 10 years in the crypto compliance functions? Yeah. So, let's take let's take both of those. So, I actually really do think um you'll have a dramatic reduction in false positives. Um and the reason is just um you know, as these tools and models get smarter, we've got so much data that we'll really be able to sort of pattern recognize. Uh and look, when Binance, we've got 300 million users, right? Um we will the models will get trained and we'll be able to say with really a high level of certainty that this pattern even notwithstanding the fact that this has alerted, we have seen this type of user, this type of activity millions of times before and we know with a a high degree of certainty uh a certainty that will still that will satisfy regulators and will satisfy ourselves that this can be closed without human intervention. So, that is the goal. And obviously, we'll we'll we can come back to that. I mean, there's got to be a lot of sort of validation before you get to that point, but I think we will. Um uh and then again, that allows the humans to now uh look at the more sophisticated, the more the more novel schemes. Um you know, you mentioned in the intro, you know, I I have experience prior to coming into finance um on the law enforcement side. I was at DEA and then I was at a pros- I was a prosecutor, federal prosecutor before that. Um and I'm well aware uh criminals, bad actors will get uh as sophisticated uh you know, we will it's a game of cat and mouse. Uh and they'll come up with new new ways and new mechanisms. So, I'm under no illusion that AI will solve everything, but it just allows it allows the humans to sort of focus on where the on where the new challenges will be.
Original Description
Interview with Noah Perlman, Chief Compliance Officer of Binance.
We explore the future of crypto compliance and the evolving role of humans as artificial intelligence becomes more deeply integrated into detection and monitoring systems.
🔹 Why false positives in compliance are expected to decrease significantly
🔹 How AI models are trained to recognize patterns with high levels of certainty
🔹 Which alerts could be resolved without human intervention while still meeting regulatory standards
🔹 Why validation remains critical before fully autonomous AI decision-making
🔹 How human experts will shift toward investigating more complex and sophisticated threats
🔹 Why compliance remains a constant “cat and mouse” game despite advances in AI
As AI takes over more routine processes, the role of humans in crypto compliance is not disappearing — it is evolving toward higher-value, more strategic work.