Artificial intelligence and dipfakes
The first wake-up call regarding security and artificial intelligence (AI) issues was the proliferation of so-called dipfakes. This was announced by the head of security of the cryptocurrency exchange Binance Jimmy Su. In his view, the proliferation of dipfakes will intensify following the software used by artificial intelligence.
A dipfake itself is an artificially created image or video of a person. Existing pictures or clips are superimposed on each other and a new product is created, the realism of which is very high. It would be difficult even for AI professionals to distinguish voice, facial expressions or body movements from real ones;
Naturally, attackers couldn’t avoid such tools. For example, in the crypto industry, fraudsters use dipfakes to try to pass themselves off as well-known industry representatives.
However, synthesized images and videos can also play a positive role. For example, if you need a photo urgently and there isn’t one.. Or you need to be in the zoom and doing something else at the same time – there are many options. But the use of dipfakes by crooks is a very different scenario.
The case of Sam Bankman-Fried, for example. In November 2022, someone under his verified account using special software released a video clip in which the former CEO of FTX allegedly promises to compensate users of the exchange. The video went viral quickly enough on the Internet. In fact, it offered to click a link, after which the cryptocurrency wallets of gullible users would be emptied.
According to Jimmy Su, the only way to combat dipfakes is to educate users about the miscalculations that AI makes when it tries to make a new image or video.
Dipfakes and KYC
Representatives of the cryptocurrency exchange Binance, including the site’s head of security, believe that the most vulnerable place for dipfakes to attack is the KYC (“know your customer”) identification system. For example, users’ personal information and finances can be compromised.
As of 2023, KYC is used by almost all centralized exchanges (CEX). In this regard, this system will be threatened not only in Binance, but also on other platforms. How will this play out?
The principle is as follows:
- First, one searches the internet for photos to create a new product.
- Using the images you find, you create a video.
- Newly created characters are published that are hard to distinguish from the real thing.
Jimmy Suh believes that mostly dipfakes are used to spread all sorts of scam in crypto-enthusiast communities. At the same time, in his opinion, technology has not yet reached a level where it is impossible to distinguish between the virtual and the real. He said:
.
“Some verification processes ask the user to blink their left eye or look left or right, down or up. Dipfaces are now quite advanced and can perform these commands. However, when we look at these videos, there are certain parts that we can detect with the human eye, such as when asked to turn our head sideways.”
Binance has already started publishing information on its blog about how to properly manage the risks associated with dipfakes. Exchange representatives admit that they themselves use AI and machine learning to detect unusual patterns in logins and transactions, as well as other “anomalous” phenomena on the platform.
In this way, artificial intelligence is being used by two parties at once. It can both be harmful, if used by crooks, and beneficial, protecting against dipfakes.
A Web3 security specialist nicknamed Noerek.eth wonders if dipfakes could be used in Ledger’s “Recover” update, because if you lose your cid phrase, you can request it by pre-verifying it through KYC.
“What if someone hacks your Ledger and registers on your behalf using their KYC? If someone uses an AI generated fake ID or dipfake to recover from companies? Twitter user @zachxbt actually made a fake KYC ID and successfully joined a couple of cryptocurrency exchanges, just like ragpolls and scammers do to withdraw their money. They didn’t divulge the sideline to the public, but they showed a new direction of attack on how they can be recovered and join the platforms.”
.
The use of AI by cryptocurrency platforms
More and more cryptoprojects are trying to integrate artificial intelligence technology. There are different ways of doing this. Very often platforms mention AI just for marketing purposes, while there is nothing fundamentally behind the projects. The schemes used are quite simple. More often than not, it’s a reference to a name you’ve heard of, such as ChatGPT or AI.
However, this does not prevent fundamentally strong projects from. For example, 4 cryptocurrencies from the “Top AI & Big Data Tokens” section are in the top 100 by capitalization according to CoinMarketCap. Projects such as The Graph (40th), Render Token (50th), Injective (69th), and SingularityNET (95th). In total, the capitalization of this segment of cryptocurrencies exceeds $5.4 billion.
Of course, it is too early to say that artificial intelligence has taken over the cryptoindustry – there are too few applied projects, if we do not take into account fraudulent ones. But already now the capitalization of these tokens is more than that of Binance USD (14th place). So there is clearly potential for development.
Chatbots, automation of trading systems and big data discovery are certainly important and will be one of the driving forces by which the next bull markets will evolve. But we still have a long way to go and we need to build better products to truly “revolutionize the industry.”
This material and the information in it do not constitute personal or other investment advice. The views expressed herein are those of the author, research portals and experts and do not necessarily reflect the views of the publisher.