Beating the Cheats: Nigeria's Tech-Driven Fight for Betting Integrity

The sports betting industry in Nigeria is facing a significant challenge with the rise of fraudulent activities, such as match-fixing, insider trading, and illegal betting operations, which threaten the integrity of the industry and erode players' trust.

Artificial Intelligence (AI) presents a powerful solution to combat this issue by leveraging advanced machine learning algorithms and predictive analytics to analyze vast amounts of data and identify anomalies and potentially fraudulent activities.

The relevance of AI-powered fraud detection and prevention in Nigeria cannot be overstated, as the country's sports betting industry is largely unregulated, making it vulnerable to unscrupulous operators and illicit activities. Implementing AI solutions can help restore confidence in the industry, protect players' interests, and ensure a fair and transparent betting environment.

SportsBoom.com sat down with two experts who consult for the top sports betting platforms in Nigeria for a chat on this crucial matter. Co-founder & CEO of Zero Complex AI, Kehinde Olateru and Gaming Compliance Expert & Founder of Space Universe Tech Hub, Tunji Idowu. 

The Need to evolve at the Pace of Fraudsters

As the CEO of an AI company specializing in fraud detection, can you share your insights into the unique challenges faced by the Nigerian sports betting industry in combating fraudulent activities?

“There are various challenges, one of the most prevalent is people trying to gain the system. For example,  you have people trying to see if they can go around to pay for their predictions, they want to use just one amount to pay for a particular game, but their intention is to cash out on that prediction multiple times,” Olateru told betting experts SportsBoom.com.

“Another prevalent challenge is underage betting. There are also numerous cases where people try to bypass the Know Your Customer (KYC) system, so as to make use of fictitious accounts.  When it comes to fictitious accounts, they are opened in order to use real people's details without their consent.”

What are the key data sources and features that ZeroComplex AI leverages to train its machine learning models for identifying anomalies and potential fraud in the sports betting domain?

“I can’t share every single one. But we look at patterns, location and the rate of occurrences. Those patterns lead to us flagging certain individuals as suspicious and our models have no choice but to get better, because these frauds are unrelenting in their activities,” said Olateru.  

“We’ve seen cases whereby people save money in their phone wallets and end up joining Telegram groups, which are eventually hacked into, and they end up being robbed of their money. There are regions in Nigeria where once you have suspicious accounts coming from those regions, they are flagged by us. We also have fictitious names especially in the case of KYC bypassing.”

“Another prevalent issue is bet farming.  Just imagine a situation where 1000 accounts have been opened from one Internet Protocol (IP) address- in such cases we make concerted efforts to combat that. It’s a situation where frauds hit an organization or platform from multiple devices while masking their IPs in most cases. So, it's important to put the right infrastructures in place to handle that.”

Can you walk us through the process of how ZeroComplex AI's fraud detection system operates, from data ingestion to alert generation, and how it adapts to new and evolving fraud patterns?

“To start with, it’s AI, so the algorithms keep learning and at the same time, human behavior is not static, as people will always look for ways to gain the system. In addition to that, frauds use malicious links and phishing mails to hack into people’s accounts in order to steal their credentials. For instance,  EURO 2024 is taking place and let’s say I’d like to make some predictions. However, I make a poor judgment and proceed to click on a malicious link on my phone and before I know it, my bank account has been hacked. This is a very familiar scenario which frauds engage in,” Olateru added. 

“It’s also worth mentioning that we’ve noticed that frauds tend to target the betting platforms that are relatively unknown and there are certain periods when fraud activity is much higher. For instance, once the games like La Liga and championships resume,  there will be an increase in fraudulent activity. However, betting is broad, so we do our part to keep training the models and just as the frauds get better, we also get better.”

You mentioned that you see more fraudulent activities in the upcoming betting platforms as opposed to the platforms which are more established?

“The established platforms make use of infrastructure that combats fraud on that level which enables them to pick up on suspicious activities, as opposed to what is the case for some relatively unknown platforms. I can also tell you for certain that some betting platforms are set up fictitiously for the purpose of harvesting information of unsuspecting users,” Olateru reiterated.

“We need to realize that the temptation to win big and cash out is very alluring and tempting to many people. In such cases, an unsuspecting customer sets up an account on the relatively unknown sports betting platform, the platform then sends a link to the customer to sign up. The customer clicks on the link and before they know it, their activities and banking details are being monitored remotely and then their money is gone.”

“This remote monitoring scenario which I just narrated is prevalent on platforms like Whatsapp and Telegram from which funds are also solicited. Such monitoring also enables financial fraud because such customers have basic level KYC. I don't want to mention names of financial institutions that have porous KYC processes, but some of them are being used to store value meagerly.”

In your experience, what are the most prevalent types of fraud that ZeroComplex AI has encountered in the Nigerian sports betting industry, and how do your solutions address these specific threats?

“Using betting wallets to commit financial fraud, phishing, sending malicious link and opening malicious telegram groups. The most sophisticated is bet farming which is much more coordinated and involves a lot of more people.”

How does ZeroComplex AI's approach to fraud detection differ from traditional rule-based systems, and what advantages does it offer in terms of accuracy, scalability, and adaptability?

“First, it's AI based and contextual- hence we factor in local nuances while maintaining international standards. We can scale depending on usage and traffic and can contrast. So, that's pretty much it. It's also easy to deploy in the sense that you don't have to ‘tear down your infrastructure’ and there is always feedback. It’s constantly improving.”

As an AI company operating in Nigeria, what are the challenges you face in terms of data availability, quality, and privacy concerns, and how does ZeroComplex AI address these issues?

“The majority of our data is not structured. Education is another challenge. People still don't get it. Most people think of AI as Chat GPT and there is another faction which views AI as the thing that is out to take their jobs, so there is a bureaucratic pushback. But, back to the issue of data, we simply don’t have enough data to train our AI models and the data which is available is far from structured in the sense that it can’t be used easily.”

How Do Betting Companies Tackle Fraudsters?

So, how are you working around these challenges?

“We work with organizations to help them realize that we are there to help them structure their data while respecting the privacy of the users. There is also increased awareness because AI is here to stay. Hence, we have increased partnerships with the public and private sector in terms of investing in affordable data centers,” Olateru added.

“Also, ZeroComplex has done a lot when it comes to working and generating policies pertaining to data use, privacy, collection, separation, and the likes. So, there has been tremendous help and stakeholder engagement in educating everyone to see the need for data to be collected and stored in a way that can be used to generate insights.”

Looking ahead, what are the future developments or advancements in AI technology that ZeroComplex AI is exploring to further enhance fraud detection and prevention in the sports betting industry and beyond?

“Investing more in data collection and data points. We are also looking at building better models at the infrastructure level. Those are the critical areas that we’re looking at right now to ensure that we stay ahead and have an ideal situation where AI is training AI to get better,” Olateru concluded. 

The Growing Need for a Central Monitoring System (CMS)

In terms of future developments and advancements, Gaming Compliance Expert & Founder of Space Universe Tech Hub, Tunji Idowu, seems keener on technology being deployed from the regulatory standpoint which would entail the setting up of a CMS, that would be plugged into the AI models, as practiced in countries like Senegal.

As a compliance manager, what are the major types of fraud and illicit activities you have encountered in the Nigerian sports betting industry, and how have they evolved over the years?

“Part of the service that we provide as a gaming compliance advisory to our customers is fraud resolution. In fact, fraud resolution used to take up most of our time in the early days. This was due to the lack of technology in terms of KYC verification platforms or tools which needed to be integrated into the gaming software,” Idowu told SportsBoom.com.

“Right from the onset, it was clear to us that Nigerian players or bettors are very savvy and we caught wind of this by virtue of some of the fraud resolution cases that we had to deal with, such as creation of multiple accounts. For instance, platforms which offer a welcome bonus for first time users, experience recurrent cases of the creation of multiple accounts.”

“When it came to the use of such means of identification, those were weak forms of identification that were put to use. We also witnessed numerous cases of fraudulent funding of players' accounts. For instance, if relatives of some bettors gained access to their banking information, such could be used to fund their own wallets through the use of third-party accounts. These were rampant at the early stage of sports betting, as well as the issue of chargeback fraud.”

“To a large extent, such fraud is still evolving and manifesting in the forms of money laundering, as confirmed by the National Financial Intelligence Unit (NFIU). In such cases, funds are disguised and put into wallets which can come from anywhere and by the time such funds migrate from the customer’s wallet to their bank account, it looks like clean money.”

Can you share any specific AI models or algorithms that have shown promising results in flagging potential match-fixing, insider trading, or illegal betting operations in the Nigerian context?

“While most of these AI solutions come as part of the package offered by platform providers, the algorithms used are usually similar. Algorithms like Random Forest which is a popular model for using multiple data points to come to a decision, which means the algorithm makes use of all the information provided to it from multiple sources to come to a single decision. For example, to categorize a payment as potentially fraudulent, the model identifies suspicious patterns and behaviors that may indicate fraudulent activities based on the data it has been trained on,” Idowu added.

“Another common model used is Graph Neural Networks (GNNs). These networks excel in identifying and analyzing complex relationships and interactions within an operator’s platform, which can be crucial for uncovering collusion or coordinated fraud. These are the 2 major models in Nigeria, but of course, other AI models are being developed in terms of a holistic platform, but those of us participating in the creation of these new models are looking at it from the regulatory standpoint.”

Can you share any specific AI models or algorithms that have shown promising results in flagging potential match-fixing, insider trading, or illegal betting operations in the Nigerian context?

“While most of these AI solutions come as part of the package offered by platform providers, the algorithms used are usually similar. Algorithms like Random Forest which is a popular model for using multiple data points to come to a decision, which means the algorithm makes use of all the information provided to it from multiple sources to come to a single decision. For example, to categorize a payment as potentially fraudulent, the model identifies suspicious patterns and behaviors that may indicate fraudulent activities based on the data it has been trained on,” Idowu added.

“Another common model used is Graph Neural Networks (GNNs). These networks excel in identifying and analyzing complex relationships and interactions within an operator’s platform, which can be crucial for uncovering collusion or coordinated fraud. These are the 2 major models in Nigeria, but of course, other AI models are being developed in terms of a holistic platform, but those of us participating in the creation of these new models are looking at it from the regulatory standpoint.”

What Role Should Regulatory Bodies Play in Promoting AI Solutions when Combatting Fraud?

Speaking of regulations, what role should regulatory bodies like the National Lottery Regulatory Commission (NLRC), or the Nigerian Gaming Commission play in promoting the adoption of AI solutions for fraud prevention and ensuring a level playing field in the industry?

“The oversight role is in place, but from what I’ve observed based on my experience, it’s simply not enough- technology needs to be deployed from the regulatory standpoint. Within the Nigerian regulatory ecosystem, we have the NLRC, Federal Inland Revenue Service (FIRS), Nigeria Data Protection & Privacy Regulations (NDPR) and NFIU that regulate the financial aspect. Within these organizations, they should be able to deploy solutions that automate compliance requirements,” Idowu commented.

“The reality is that more often than not, the solution to these issues come from the operators, not the regulators. That’s because the operators are trying to ensure that they comply with the requirements whether it has to do with AML or terrorism financing. For instance, we see the efforts of the operators in terms of KYC processes by trying to ensure that they are not only able to collect data but verify it in terms of data audit of all the data subjects.”

“But, back to the regulators, when it comes to fraud issue reporting, pertaining to the responsibility of the NFIU, a CMS should be able to capture all these reporting models. It’s for this reason that the regulators must be able to deploy a CMS that verifies the veracity and the accuracy of the information that the AI models are verifying. This is highly imperative because we have seen cases where an AI model is trained to detect fraud, identify patterns of fraud and prevent it, but there are concerns about what other instructions have been given to these AI Models outside of that, which may affect data processing, collection and governance.”

“This CMS would need to be plugged into all these AI models. We see such practices in the UK and Senegal, but in Nigeria’s case, the issue of a CMS has been muted for the longest time because the quantum of data that’s being collected is massive and the data integrity and security of Nigerians are placed in the hands of foreign entities which are not localized in Nigeria.”

“Most of the gaming software providers are foreign owned and not domesticated in Nigeria, which automatically means that they aren’t subject to data audit in Nigeria. Therefore, the regulators must be able to develop their own CMS that could at least capture the data, monitor the data use and be able to speak to all the AI models that they are using for fraud prevention. That way, the industry is winning in terms of the operators being satisfied with the integrity upon which they operate, and genuine players gain confidence that everyone is on a level playing field.”

Future Advancements in AI Technology in the Sports Betting Industry

Looking ahead, what are the future developments or advancements in AI technology that you anticipate could further enhance fraud detection and prevention capabilities in the Nigerian sports betting industry?

“Deployment of an AI model KYC collection verification tool, detection of money laundering in terms of disguising wallet funding into customers’ account as a genuine inflow when it’s not. Another one is the data audit platform that is able to really help to monitor the data collection processing and audit.”

“In addition to the above, I would like to see the acquisition of a revenue assurance platform that is able to automate revenue collection. In the gaming sector, we have about 1.3 billion dollars that is leaving the country monthly. From that amount, the government needs to be able to collect accurate tax via the automated revenue assurance platform, which will also enable the optimization of revenue collection for the government. Individuals or companies that are able to come up with these kinds of AI models that will lead to improvement on the entire market, will certainly lead the pack,” Idowu concluded.