How customer screening for Anti Money Laundering (AML) works

On March 26, Binance together with IdentityMind announced the launch of an AML service. On February 21 of the same year, Huobi Global issued a press release about AML policy tightening. And four of the biggest Korean banks earlier established a hotline where users can share information on any illegal activity related to crypto assets.

We can now observe that the crypto market responds dynamically to the pressure of the state institutions and aims to stay away from suspicious and risky transactions. And although illegal businesses have recently favored the use of crypto assets as their payment instrument*[1], blockchain can evaluate exactly how close these transactions are to illegal activity, and do that almost completely automatically.

The reasons why.
1. Nothing is ever deleted on the blockchain
Once a record has been added to the blockchain, it can never ever be deleted.
The blockchain is an uninterrupted chain of linked entities
Every block contains information about the previous blocks.
2. Database can never be lost
Since blockchain is a distributed technology, a copy of the database is stored on computers of its every user. We are talking about millions of users here. And since nothing is ever deleted on the blockchain (see par. 1 above), millions of users would have to be reached to delete incriminating evidence.
3. Every transaction has unique references
With transaction address (TxID) every amount of transaction, the source of assets or destination can be backtraced.
4. Crypto assets are complex
Every Bitcoin asset in your wallet consists of smaller parts.

Summing it up, with analytical algorithms all assets can always be backtraced. If a transaction came through a risky address А, then no matter how long the chain to the destination address Z is, Z’s relation to A will be revealed.

How so. A Bitcoin address belongs to a legal entity or to an individual entity (let’s call them ‘the entities’). The algorithm examines public and private sources, like forums or national registers, and labels the entities. Categories of labels include:
• darknet platform
• verified exchanges
• totalizators and gambling games
• malware
• trading platforms
• miners
• mixers, blenders
• exchange offices with weak KYC

and so on. Every category is evaluated for risk. For example, darknet is an entity with high-risk value, whereas exchange offices with weak KYC are entities with medium risk value.

One entity can have multiple addresses, and one entity can generate an infinite number of transactions. Once the algorithm labels an entity, it tracks it and labels all addresses and transactions linked to that entity.

A transaction or an address can have multiple links, a transaction often has many inputs and many outputs, and infinite number of transactions can pass through a wallet address. This is an example of a random transaction with over 100 inputs and outputs, and this is only the beginning of the list.

It’s merely impossible to analyze big data arrays manually, but even a computer program has to be built upon some algorithm which can properly process all the links. The graph-based algorithm is a solution. Here, wallet addresses (entities) represent the vertices and Bitcoin values sent between those addresses represent the edges. The algorithm labels each vertex and uses the links between them to send risk evaluations to target addresses.

In the picture above, the addresses without any direct or immediate link to the “bad” entities, like Darkmarket, Gambling, have low-risk value. Whereas addresses G, H, I and J have high-risk value because of the direction of intermediate links to the labeled vertices.

The most interesting thing is the primary entity labeling. As was mentioned above, the information for labeling is gathered from open sources and private databases which store long lists of entities involved in illegal activities. A smart algorithm extracts information from these sources to fill its own inner database with expert data about entity labels. The expert database is being enriched with new labels 24/7. If we reload this graph over every 1–2 weeks, we can quickly get risk values for target addresses.

The point is, such AML screening software protects the assets and reputation of its crypto owners on an automatic basis. And I advise being careful with transactions coming into your wallet to avoid being blocked on the exchange or experience any other unwanted troubles.

*[1]According to the Europol report

CEO AMLBot/AMLSafe, here to help you be compliant