Congratulations, KYC Is Now a Vibe Check

[Josefjix]

For me, the best solution for exchanges should be something not complicated. AI bot asking some set of predefined question by the user should be just enough, I see no need for the facial and video verification.

Yes, the same thing Social Media platform doing to check and track out any illegitimate videos by content creators, uses AI bots to detect copied or plagiarized content thereby reducing rewards on the any creator who violate that. The same thing crypto exchange should do to detect any kyc done via AI generated images, it could literally help the system to filter out many fake sign ups and more.

[ZAINmalik75]

https://x.com/venturetwins/status/2011285029541077033

Now that AI is capable enough to accurately mimic anyone if you have a camera, a laptop and an Internet connection, how reliable are current Know Your Customer measures? They aren't. So let's give some feedback to the exchange companies: What's the next measure for the exchange to reliably verify your identity? Go vote.

It's as if Know Your Customer is still put in place, not to prevent money laundering, but to push everyone to voluntarily hand over their bitcoin activity to the government. I know, a wild conspiracy theory that I read on X.

KYC without face verification is a joke, obviously, because AI is now capable of creating almost anything especially if it's in image form, videos could still be somewhat understandable that they have been created using AI, but with images, sometimes the perfection is so good that one could never guess whether it's a real image or a fake one created using an AI image creation model. This is why if a platform doesn't ask for facial verification when asking their customers for KYC then they are definitely not doing it right.

I use Binance most often, and it asks for facial verification even when you have completed the initial verification, especially when you are doing P2P trades sometimes, or even making a withdrawal, etc., and when they ask for facial verification, you are asked to blink your eyes, nod your head, and look left and right, and you pass the test only when you follow the instructions correctly, and I don't think AI can do that just yet. What's shown in that post could be done using an App or something, but when you phone's camera opens directly, it won't be able to do that for you.

[NotATether]

A lot of KYC checks are really unnecessary for the vast majority of users doing swaps and exchanges, or simply moving money from Point A or point B, but the ones who want an official presence need to get permits and licenses and one of the requirements is through AML procedures.

Now there are obviously many crypto companies with lax and outright terrible AML mismanagement within their own company & funds, but it's easier to simply pass the blame to the customers ("Our users were trying to launder money." Do you know how ridiculous that sounds coming from a spokesperson? ) and take the fine than to spend even more money to fix their own internal structures.

An exchange like Coinbase doesn't have this problem because they ask for so much personal information, which I understand makes people here resentful of that platform, and cross-check it that it's almost impossible to fake it with AI alone. But the rest of these platforms are cooked.

[Outhue]

I am quite lost here, is this all about mimicing someone appearance and get access into their account? Or using AI to pass KYC verification? Which one exactly? I still find it hard to believe that AI can pass facial recognition from the likes of Binance because those verification are lIve.

The AI needs to be able to come in human form and hold a ID for the Live camera to take full effect, tell me how AI is going to do this or there is something that I don't just get about all these?

[PrivacyG]

The thing is.  It was always like this and who pushed for the Know Your Customer general implementation knew this would happen.  Before Artificial Intelligence, some body with great Photoshop skills could fake it well.  Technology evolving and ultra realistic generated images before all of this, it all showed signs that things may get dystopian soon.

More over.  I have very big doubts that Governments did not know Artificial Intelligence was coming.  I am guessing AI was worked on for a long while before it was actually publicly released.  You pretty much only needed a brain and a little bit of critical thinking to realize where this was going.  I hope I am not wrong when I say this but I am sure Meta previewed Virtual Reality communication that scans your face and replicates it so you could communicate with someone from a distance while looking at a 3D version of them moving before AI was a thing.  Microsoft had that gigantic Augmented Reality headset that tried to replicate your voice and person.

I repeat.  Anyone with a brain could have seen where this was going.

And then what is the point of Know Your Customer if this was the predictable future?  Because the more I think about it the more I realize Know Your Customer is only a bridge to accommodate to much more invasive procedures such as retina scans and finger prints.  People are already used to handing over their finger print to a device, fully Trusting a corporation and the code that lays behind that fancy User Interface.  Why would they not hand over a finger print to have a Meta account?

It keeps getting scarier however.  You can see how things worked out for The United Kingdom and their hard Censorship.  The hacks, the leaks, the bans, every thing only double and triple confirmed that my paranoia was never for nothing and that when a Government CAN do more and the society will NOT oppose, they will test out the limits of society in a very eerie way almost like an experiment on humanity.

What is next?  As I predict above, probably finger prints and retina scans next to Digital IDs.  And when the retina and finger prints of millions of people get leaked and all the visual unique identifying methods are used and compromised, the only thing left is to verify who you are by DNA.

[BlackHatCoiner]

What is next?  As I predict above, probably finger prints and retina scans next to Digital IDs.

Jokes asides, I don't think there's anything next. AI will help criminals compromise accounts, and the government will just let it be, because reverting to non-KYC relinquishes their enormous privilege to spy anyone with a cheap excuse. They will simply try to regulate AI, which will inevitably lead at some point to AI companies requiring KYC for software that can be used for this purpose, such as deepfake. (Which will obviously be bypassable by criminals, as with piracy, AML and tax evasion.)

More Know Your Customer incoming!

[Pablo-wood]

Every citizen has a valid means of identification issued by the government and those IDs have tracking numbers so I feel emphasis might dwell more on the IDs and how it matches the selfies and video verification sent be it AI generated or real.

If the case is barely KYC I think there is still no serious threat to the centralised system. As for theft cases where imposters might want to gain access into users account I believe IDs and live face capturing might mitigate this situation.

[Furball808]

As funny as it sounds, the first one may actually be the most convincing one. AI is capable of giving answers with some emotional depth already so existential questions can be answered. Your pets doing taxes? AI can make one and even more. Literally AI will make it harder for everyone to verify each other’s identities. It should be illegal nowadays to spread an AI generated picture without specifying it is AI.

[goldphysicalbitcoin]

In the future, as AI advances rapidly, not only will KYC become increasingly unreliable, but all identity verification methods based on images, video, or voice will lose credibility.

Remote identity verification will no longer be able to rely on personal traits such as facial features or speech patterns. Instead, cryptographic proof — particularly identity binding through blockchain-based private keys — may emerge as a more robust solution.

The reason countless people have failed to impersonate Satoshi Nakamoto is simple: they do not control the private keys.

[AbuBhakar]

Every citizen has a valid means of identification issued by the government and those IDs have tracking numbers so I feel emphasis might dwell more on the IDs and how it matches the selfies and video verification sent be it AI generated or real.

If the case is barely KYC I think there is still no serious threat to the centralised system. As for theft cases where imposters might want to gain access into users account I believe IDs and live face capturing might mitigate this situation.

In fact, centralized exchanges are already doing this now. Apart from ID, there are also video verifications that require movement that they themselves will provide. Even AI might be difficult, but obviously, it will improve even more eventually, especially because AI has improved so much now. Apart from IDS, they might also put in codes that only the owner will have.

As funny as it sounds, the first one may actually be the most convincing one. AI is capable of giving answers with some emotional depth already so existential questions can be answered. Your pets doing taxes? AI can make one and even more. Literally AI will make it harder for everyone to verify each other’s identities. It should be illegal nowadays to spread an AI generated picture without specifying it is AI.

AI is so advanced today that even those in the medical field are considering using it because it can produce faster results, sometimes even more accurate. I know people in the medical field who are also studying AI now, so they can use it in the future.

[hero_the_bossman]

Every citizen has a valid means of identification issued by the government and those IDs have tracking numbers so I feel emphasis might dwell more on the IDs and how it matches the selfies and video verification sent be it AI generated or real.

If the case is barely KYC I think there is still no serious threat to the centralised system. As for theft cases where imposters might want to gain access into users account I believe IDs and live face capturing might mitigate this situation.

In fact, centralized exchanges are already doing this now. Apart from ID, there are also video verifications that require movement that they themselves will provide. Even AI might be difficult, but obviously, it will improve even more eventually, especially because AI has improved so much now. Apart from IDS, they might also put in codes that only the owner will have.

I think the systems that check the videos will evolve with the AI itself.

Nothing is forever in this world and new ways change the old ones, in spaces like crypto or AI - even faster than we may see it.

[sokani]

Is there any good in the creation of such a tool? All I can think of are its fraudulent uses. With such a tool, scammers can easily impersonate celebrities and scam their victims during video chats, and they can also pass KYC checks on exchanges. Just like we have tools that can detect AI-generated texts, I hope developers can build algorithms that will be integrated on exchanges that can help detect AI-generated images and videos used for KYC because this shit is scarier.

[Satofan44]

I am quite lost here, is this all about mimicing someone appearance and get access into their account? Or using AI to pass KYC verification? Which one exactly? I still find it hard to believe that AI can pass facial recognition from the likes of Binance because those verification are lIve.
The AI needs to be able to come in human form and hold a ID for the Live camera to take full effect, tell me how AI is going to do this or there is something that I don't just get about all these?

Don't write this spammy shitpost and make false claims here. AI is used trivially to bypass live verification. Stay in threads that are more appropriate to your limited knowledge and brain functionality.

The thing is.  It was always like this and who pushed for the Know Your Customer general implementation knew this would happen.  Before Artificial Intelligence, some body with great Photoshop skills could fake it well.  Technology evolving and ultra realistic generated images before all of this, it all showed signs that things may get dystopian soon.

Photoshop being required was already considered an advanced implementation of this. In the most early versions of it, I have successfully bypassed it using a static image. It is laughable how rudimentary and stupid these systems are. Most of the security measures that they implement are obfuscations, they are not real measures. Some exchanges may require uses to explicitly use a smartphone to do video verification, so that they prevent people plugging in a software feed to the camera input. Of course, this does not accomplish such thing. It merely makes it slightly harder. You can create a fake camera on smartphones too and stream data to it, therefore you can either provide images to it, replay a prerecorded video or an AI deepfake.

More over.  I have very big doubts that Governments did not know Artificial Intelligence was coming.  I am guessing AI was worked on for a long while before it was actually publicly released.  You pretty much only needed a brain and a little bit of critical thinking to realize where this was going.  I hope I am not wrong when I say this but I am sure Meta previewed Virtual Reality communication that scans your face and replicates it so you could communicate with someone from a distance while looking at a 3D version of them moving before AI was a thing.  Microsoft had that gigantic Augmented Reality headset that tried to replicate your voice and person.

I repeat.  Anyone with a brain could have seen where this was going.

You are confusing incompetence with lack of information. The governments definitely knew that AI was coming, I have been warning about the potential for deep fakes for 2 decades now. They are not that blind, they are just often quite incompetent.

What is next?  As I predict above, probably finger prints and retina scans next to Digital IDs.  And when the retina and finger prints of millions of people get leaked and all the visual unique identifying methods are used and compromised, the only thing left is to verify who you are by DNA.

None of this will work at all. Those data points are equivalent to usernames, and nothing prevents me from harvesting usernames on a large scale and using it to access accounts. Anyways, you have no idea just how many fake citizens exist in the world. I am talking about completely valid, government issued birth certificates for people that do not exist. You can just use the DNA of young people who died and pretend that it is your DNA. There are many ways to bypass this, it would be a joke that is worse than the current system.

[fikrett]

Is there any good in the creation of such a tool? All I can think of are its fraudulent uses. With such a tool, scammers can easily impersonate celebrities and scam their victims during video chats, and they can also pass KYC checks on exchanges. Just like we have tools that can detect AI-generated texts, I hope developers can build algorithms that will be integrated on exchanges that can help them detect AI-generated images and videos used for KYC because this shit is scarier.

Yep. There are 100% patterns that AI won't be able to make a replica for.

I hope for your scenario too.. But AI will keep becoming better and better, if this bubble won't burst one sunny day.