How AI protect mobile money users

DAR ES SALAAM: MOBILE money has become part of everyday life in Tanzania. Today, people use their phones to pay school fees, buy food, pay rent, receive salaries and send money to relatives in villages.

From Dar es Salaam to Mbeya, from Mwanza to Mtwara, mobile money has made financial services faster, cheaper and more accessible than ever before.

According to data from the Tanzania Communications Regulatory Authority (TCRA), millions of mobile money transactions are conducted daily, worth trillions of Tanzanian shillings each month.

This growth has helped small businesses, farmers, women and youth participate in the digital economy. However, as mobile money continues to grow, fraud has also increased.

Most mobile money fraud in Tanzania does not involve hacking computer systems. Instead, criminals target people.

Real-Life Examples of Fraud in Tanzania The described fraud methods are not theoretical. Tanzanians have fallen victim to sophisticated schemes:

• Fake SMS/Prize Scams: A prevalent scam involves messages claiming the user has won cash or airtime from a popular show (e.g., Bongo Star Search) or a ‘loyalty reward’ from their mobile network operator (MNO).

Recipients are directed to a link or to call a number to ‘claim’ the prize, leading to phishing or social engineering.

In 2020, the Bank of Tanzania (BoT) and the TCRA issued a joint public warning about a surge in such fraudulent messages, especially those impersonating banks and MNOs.

• SIM-Swap Fraud: This is a critical threat. Criminals corrupt telecom agents or use forged IDs to illegally swap a victim’s SIM to a new card.

They then request mobile money PIN resets, gaining full control.

A high-profile case in 2022 involved a criminal ring in Dar es Salaam that used insider collusion to execute SIM-swaps, draining millions of shillings from victims’ mobile wallets before being apprehended.

• Social Engineering Calls (‘Vishing’): Impersonators call, posing as customer service from M-Pesa, Mixx by Yas or Airtel Money.

They often claim the user’s account is ‘compromised’ and urgently needs verification, tricking them into revealing their PIN.

Others pretend to be relatives in distress needing immediate funds. AI in Action: Tanzanian Case Studies and Best Practices To fight these threats, mobile money providers now use Artificial Intelligence systems.

AI does not just look at individual transactions. It studies patterns of behaviour over time.

It learns how much a user usually sends when they send money, where they are located, and which device they normally use.

AI defence systems are already active on Tanzanian platforms.

Their design follows international cybersecurity standards like the ISO 27001 framework for information security management and the Payment Card Industry Data Security Standard (PCI DSS), adapted for mobile financial services.

• Detecting Anomalies: If a user in Mbeya who typically sends small amounts between 8 AM and 8 PM suddenly attempts to transfer 500,000/- at 2 AM from a new device, the AI’s risk engine scores this as high-risk.

The transaction may be held for additional customer verification (e.g., a one-time PIN via USSD or a callback).

This real-time behavioural analytics model is a Financial Action Task Force (FATF)-recommended practice for mitigating transaction fraud.

• Network Analysis vs Fraud Rings: AI excels at connecting dots. For example, one of the MNO’s in Tanzania, has publicly discussed its use of AIdriven Fraud Detection Systems (FDS).

If multiple new accounts are opened from the same device in a short timeframe and immediately receive small test transactions from a network of other accounts, AI can flag this as a potential mule account ring used to launder stolen funds. This pattern-based detection is far faster than manual human review.

• Learning from Attempts: When a new scam emerges (e.g., a wave of phishing messages with a specific keyword), AI models learn from the first reported incidents.

If thousands of similar messages are then sent, the system can automatically block them or warn recipients.

This adaptive machine learning is a core principle of modern cybersecurity. AI against organised fraud networks Fraudsters rarely work alone.

They often operate in organised groups using multiple SIM cards, devices and accounts.

AI systems are very effective at identifying these networks by linking patterns that humans cannot easily see.

Over time, AI systems learn from past fraud cases. The more fraud attempts they detect, the smarter and more accurate they become.

ALSO READ: Tanzania leads charge as mobile money sector hits $1.1tri milestone in sub-Saharan Africa

Why Delays Happen: A Necessary Inconvenience The frustration over a delayed transaction is understandable, but it’s a sign of a protective system.

The GSMA’s Mobile Money Certification programme, which several Tanzanian providers participate in, mandates robust risk-based transaction monitoring.

A delay often means the AI has triggered a ‘step-up authentication’ protocol.

For instance, a large transfer to an unfamiliar recipient might require a biometric check if supported, or a mandatory waiting period, a practice that has saved countless users from impulsive, scaminduced transfers.

This approach follows international best practices used by global payment systems such as Visa and Mastercard, and aligns with standards like ISO/IEC 27001 and the NIST Cybersecurity Framework.

The Irreplaceable Human Role: Tanzanian User Vigilance AI cannot compensate for poor personal security hygiene. Tanzanian users must:

• Protect PINs and Personal Information: Never share your PIN or secret codes. MNOs and legitimate banks never ask for this via call, SMS or email. The TCRA consistently runs public awareness campaigns emphasising this.

• Update Registration Details: Ensuring your SIM is registered in your correct legal name (as per NIDA) and updating details upon losing an ID card prevents fraudsters from hijacking your identity for SIM-swaps.

• Report Suspicious Activity: Immediately report lost phones or suspicious messages to your MNO’s official customer care line and to the TCRA Short Code 133 (free of charge). This data feeds and improves the AI model.

Benefits for ordinary Tanzanians AI-powered fraud protection brings clear benefits to everyday users.

These include fewer successful scams, faster detection of suspicious activity, reduced financial losses and increased confidence in digital payments.

This protection is especially important for small traders and low-income users who cannot afford to lose money.

Building a Trusted Digital Economy: The Way Forward Tanzania’s digital finance journey, highlighted by the National Financial Inclusion Framework, relies on trust.

AI-driven security, aligned with BoT’s regulations for electronic money issuers, is a silent pillar of this ecosystem.

Best practices from leaders like Kenya’s M-Pesa show that continuous investment in AI, combined with relentless user education, reduces fraud losses significantly.

Conclusion

The battle against mobile money fraud in Tanzania is real, but so are the intelligent defences. From thwarting SIM-swap rings in Dar es Salaam to delaying a suspicious midnight transfer in Mwanza, AI works tirelessly.

By embracing both this technology and their own role in security, Tanzanian users can continue to leverage mobile money with greater confidence, fuelling a more inclusive and secure digital economy.

As Tanzania continues its journey toward a digital economy, trust is essential. AI-driven cybersecurity is quietly strengthening that trust by protecting mobile money users everyday.

The future of mobile money security in Tanzania is not just faster. It is smarter, safer and more reliable.