Internet Star Companies Are Involved in Money Laundering. How Can We Solve the Problem of the Continued Existence of This Black Industry Despite Repeated Bans?

The Bad Review team recently revealed that an online gambling gang used Pinduoduo as a payment channel to launder money. This has attracted a lot of attention online, but the incident has not yet been concluded.

The upgrade of payment methods has made money laundering no longer a transaction we see in gangster movies, but has kept pace with the times with the advancement of technology. However, at the same time, machine learning has brought better detection and crackdown methods to deal with the illegal operation of online money laundering.

The day before yesterday, the public account platform "Chaping" published an article, which caused quite a stir.

In this article calledPinduoduo stores have become money laundering platforms for gambling websites, with a single store laundering 500,000 yuan a day"In the article "The Bad Review Team accidentally discovered that some shops on the Pinduoduo platform were actually selling dog meat under the guise of genuine goods. They were claiming to be virtual currency recharges, but secretly they were payment channels for gambling websites to collect payments.

It turned out that the bad review team wanted to conduct research on the gray industry of online gambling, but when they were conducting a recharge test on the gambling website, they found that the payment interface of the gambling website actually jumped to Pinduoduo's payment page on its own, and finally the transaction was completed in the form of an order for a virtual product. The gambling money became the transaction amount, and the "money laundering" was completed openly.

During their investigation, they found that because Pinduoduo has a low store review threshold, there are many online stores that act as money launderers. The capital flow behind them is also astonishingly large, and can even reach a daily transaction volume of 500,000 yuan.

However, for unknown reasons, the Pinduoduo platform did not carry out any supervision.

As soon as this news came out, Pinduoduo could no longer sit still.

Soon they issued a statement, saying that the article was false and untrue information, and would sue the negative review team for slander and defamation, with a compensation amount of up to RMB 10 million.

The negative reviews immediately criticized Pinduoduo, saying that it was clear that its supervision was not strict, but it still held the media responsible for reporting it. The incident is still fermenting...

The cat-and-mouse game between money laundering and anti-money laundering

The harm of gambling is self-evident.

With the development of the times, gambling has also developed online business and has become a very important part of the online gray industry. The popularity of various online payment methods has allowed gambling, which was originally hidden in the Internet, to use all kinds of methods to improve its product experience and attract more users.

If the Pinduoduo incident that was exposed this time is confirmed, they will probably be to blame for the money laundering channel caused by the loophole.

Although, with the improvement of laws and regulations, many channels have been regulated or even banned, but there are always loopholes to exploit, helping casinos to launder money and legalizing more gambling funds and banker income has become a way for those cyber criminals to make money.

Money LaunderingThe term originated in the early twentieth century, when there was a huge criminal group in Chicago. The leader was a gangster named Alphonse Gabriel Capone. They had a large amount of cash obtained through various illegal channels, but they did not dare to deposit it in the bank.

The financial director of the criminal group bought a large number of coin-operated washing machines and started a laundry business. Every night when calculating the day's income, he added the stolen money to the list and reported it to the tax department. In this way, after deducting the taxes payable, the remaining illegal money became legal income.

Later, this method of legalizing money from illegal sources was called "money laundering."

The amount of illegal money laundering activities around the world is so large that it even accounts for about 2% to 5% of the global GDP, which is about 800 billion US dollars per year. Moreover, with the development of modern technology, the channels for money laundering are becoming more and more abundant and more and more difficult to detect.

In response to the increasingly complex money laundering methods, the international community has proposed risk-based anti-money laundering joint prevention and control regulatory measures.Identification of suspicious transactions based on rules and featuresThis requires a lot of manpower and the effect is not good.

According to a Europol report, of the suspicious transaction reports submitted by financial services institutions, approximately 10%s require further investigation by the authorities.

So what changes will artificial intelligence bring in the face of the dilemma of anti-money laundering?

How machine learning can help with anti-money laundering

The core of anti-money laundering is to do a good job in user identification: first, customer identity identification, and second, identification of the source and destination of funds, in order to determine whether the fund transaction matches customer attributes and other characteristics.

In anti-money laundering,It is difficult to process massive amounts of data and judge suspicious transactionsThe introduction of technologies such as machine learning will bring new opportunities for solving these problems.

In some cases, machine learning methods are used to learn the judgment patterns of senior anti-money laundering experts, classify and sort suspicious cases, greatly reduce the screening base, and achieve efficient human-machine collaboration.

FirstMillions of dataconductMultiple feature dimensions, and then throughMachine learning models,Achieved the identification of suspicious casesAutomatic classificationandSort.

Incorporating the experience of anti-money laundering experts into the machine learning system can help the system automatically tune and evolve. After a short period of training and tuning, the machine learning system can approach the level of senior anti-money laundering experts, saving a lot of manpower costs.

In addition, technologies such as semi-supervised machine learning also have great potential.

For example, it can be used to identify complex money laundering transactions and underground banks, using behavioral data and a small number of feature tags to identify abnormal transactions and relationship graphs through graph analysis, clustering, association analysis and other technical means. Combined with the experience of experts, it is expected to uncover those money laundering organizations hidden in the dark.

Moreover, the longer the algorithm model runs, the more cases are input, and the more human corrections are made, the stronger its ability to identify suspicious transactions will be, and the lower the probability of misjudgment and missed judgment will become.

China faces particularly great pressure to combat money laundering.

Chinese society has been particularly transformed by the abundance of Internet products. As the country with the most popular, convenient and mature mobile transactions in the world, China faces the greatest pressure to identify suspicious transactions because it faces the transaction flows of more than one billion people.

According to reports, some of the country's leading financial technology companies,Through machine learning model methods, it is possible to reach the level of senior anti-money laundering experts 95% and reduce the workload of manual review 30%, helping to effectively control money laundering activities.

In a company that provides anti-money laundering services, the solution they provide is as follows:

Enhanced anti-fraud identification at the source:Based on information such as device model, behavioral characteristics, access frequency, geographic location, etc., risk identification is carried out to promptly detect fraudulent activities such as simulators, flashing and modifying devices, and group cheating.

Build a more three-dimensional user portrait:Through the account's related information, as well as data such as credit applications, daily deposits and loans, fund transactions, and device logins, we use associated network technology to build a user relationship map, outline the user's individual characteristics, and establish a three-dimensional portrait.

Monitoring of abnormal fund transactions:Based on business data, by mining the fund transaction data contained in the relationship graph, with the help of rules or models, abnormal fund transaction behaviors and abnormal transaction groups in the graph can be identified.

Accurately locate money laundering accounts:The deep learning-based relationship graph technology helps financial institutions sort out and build retail customer relationship graphs, expands the perspectives and means of risk prevention and control, and builds a complete money laundering account identification mechanism based on relationship graphs.

This is almost the routine process of anti-money laundering technology with the support of AI technology.

Who is to blame for the lack of gray industry supervision?

Although AI is used to combat money laundering in many company reports, it must be acknowledged that AI currently mainly uses induction and synthesis rather than deduction to analyze problems.

Therefore, in current usage, the words that are emphasized the most are data monitoring, reducing manpower, etc.

Although the use of AI in anti-money laundering is an inevitable trend, the development and deployment of anti-money laundering systems is still in its infancy, and the involvement of industry experts is still a very important factor.

Technologies such as machine learning must rely on large amounts of data training to grasp some patterns of money laundering methods, thereby providing more comprehensive and secure supervision and decision-making. Perhaps the maturity of this technology will occur in the near future.

Back to the Pinduoduo incident, at present, apart from the negative reviews of this self-media, there has not been any official investigation report, so whether they are laundering money remains to be determined.

AsBad review teamAs said,

「This report is not intended to kill any particular platform, but to kill all the illegal gambling industries together."

We are the same. We hope that more artificial intelligence technologies can not only help make our lives more convenient, but also make our lives purer.