Back in November 2022, the European Comission (EC) published a report on the prevalence of money laundering in the iGaming industry, flagging it at the highest level of threat. They observed that the online gaming industry had several weaknesses regarding AML regulations and hinted at stricter regulations to come.

Coincidentally, it was around the same time when Rishi Sunak was forming the government. This amplified the rumors of looming bottle-neck regulations for the gaming industry. There was also the speculation of the UK Gambling Commission trying to implement a “single customer view” across all operators’ platforms in the country. [1]

3 weeks ago, every prominent igaming news site was covering the new AML guidelines published by the EGBA. The main aim of the report was to provide a more specific and standarized approach for AML regulations, an issue that was pointed out way earlier. Quoting news sources, the EC recommended improvements to its members. “It stated that they should lower the winnings threshold, subject to proper customer due diligence policies, to below the current level of €2,000. In addition, with the assistance of their respective jurisdictional regulators, it wants operators to ensure users can’t create multiple accounts” [2]

For online casinos and sportsbooks, it is quite a challenge to identify and flag suspicious activity or multi-accounting with such massive influx of data flowing by the hour. Operators can however, rely on AI/ML algorithms to stay compliant and ensure they effectively curb money laundering at their casino.

One model that is used across industries to ensure that users do not create multiple accounts is player deduplication’.

Deduplication is a process of removing duplicated data or records from a database or data set. This process helps to identify cases where a single player might be using multiple accounts. Fuzzy matching is the algorithm used for deduplication that compares records on the similarity of users. It makes a decision on whether they might be the same person under different aliases by using a set of rules to calculate the similarity level between records based on specific criteria. The similarity score is then used to determine if two records represent the same entity.

Similarly, an AML model can identify patterns and anomalies based on known patterns of suspicious behavior from historical data (such as money dumping) to detect potential money laundering in real-time.

Money dumping refers to a tactic where a player intentionally loses a large sum of money in a game to transfer the funds to another player in order to evade taxes or to launder money, as the transferred funds appear as winnings rather than a direct transfer. A player may launder a large sum of money by opening multiple accounts under different aliases (called ‘mule accounts’) and distribute this cash across accounts to avoid suspicion. Machine Learning becomes a casino’s best friend when it comes to monitoring and flagging such transactions that are otherwise hard to detect.

Why GAMWIT’s AML model stands out

Exclusively built for the gaming and sports-betting industry

Strong research and development backed by a leadership team with 100+ years of collective experience in gaming analytics. GAMWIT’s AML model flags risky players and segments them into high risk, medium risk, low risk or no risk buckets based on suspicious laundering activity. The model output contains relevant information such as the relationship between each predictor variable with the target variable. It also provides a Predicted vs Actual Analysis of the model outcome for maximum transparency.

Analyzes the journey of money

The model analyzes the journey of money from deposit to encashment while also taking individual player behavior into account in order to predict suspicious behavior early in the business cycle. It is often difficult to track both player behavior and the complete journey of money individually. A combined analysis of the same ensures a full picture is provided to the operator to follow the highest standards of safety and compliance.

For example, the model detects

  • Sudden unusual spikes in deposit amount in a player’s account
  • When a player deposits large amount of money and withdraws after minimum threshold spend or wagering
  • When a player’s visit frequency is inconsistent and involves large sum deposit and withdrawals
  • When a player intentionally loses large amounts of money in a game to transfer the funds to another player in a series of different games
  • These are few ways the AML model flags suspicious activity and informs the operator along with relevant insights on their individual behavior.

Country-specific regulations tracking

The model also takes into account the country-specific regulations of each player to segment the level of risk involved according to the country-specific jurisdiction. Any change in jurisdictions will be updated on the model promptly without any additional cost or complication to the operator.

To see how ML-driven deduplication or AML models work, sign up for a free demo of GAMWIT here

Also Read:

How to Leverage Machine Learning for AML Compliance