As artificial intelligence makes digital services faster and more intelligent, it is also changing the way cybercriminals target consumers. In Kenya, fraudsters are increasingly using AI-generated videos, cloned websites and impersonation tactics to trick customers into revealing sensitive information or sending money to fraudulent accounts.

Safaricom is responding by combining technology-led fraud prevention with customer education, as the telecommunications operator seeks to protect digital transactions and maintain trust in services such as M-PESA. The challenge is becoming more complex as AI tools make it easier for criminals to produce convincing fake promotions, impersonate company executives and spread misleading financial offers.

Speaking at Safaricom’s Cybersecurity Summit in October, held under the theme “Harnessing AI, Strengthening Security,” Safaricom Group Chief Executive Officer Peter Ndegwa said AI was creating new opportunities while increasing the importance of trust.

Ndegwa said Safaricom was using AI to identify suspicious activity and stop fraud before it reaches customers. According to remarks reported during the summit, the technology enables the company to prevent 94% of attempted fraud in real time by analysing customer behaviour, learning transaction patterns and identifying potential sources of risk. The figure reflects the CEO’s stated assessment of Safaricom’s fraud-prevention capabilities.

Safaricom’s fraud challenge also has an internal dimension. The company dismissed 118 employees over fraud in the year ended March 2026, according to TechMoran’s report on the dismissals. The cases highlight why AI-powered monitoring alone cannot address every vulnerability. Protecting digital transactions also requires employee accountability, strong internal controls and effective oversight.

The threat extends beyond traditional phishing messages. Criminals can use AI-generated videos, cloned websites and convincing impersonations to make fraudulent offers appear legitimate. TechMoran has reported on AI-driven cybersecurity threats in East Africa, including deepfake scams, as attackers increasingly exploit emerging technologies to target businesses and consumers. Earlier reporting by TechMoran on AI-enabled cyberattacks and the next frontier of cyber defence also highlighted the growing threat posed by deepfakes and increasingly sophisticated attacks.

These tactics may be used to promote fake giveaways, promise to double a customer’s Fuliza limit or advertise fraudulent investment opportunities. A video may appear to feature a trusted executive, an advertisement may use recognisable branding, or a message may claim that an offer expires within minutes. Customers who act before checking the source risk losing money or exposing their personal and financial information.

At the cybersecurity summit, Ndegwa emphasised that security must be built into digital services rather than treated as an afterthought. The approach places responsibility on businesses to make security part of product development, risk management and customer protection.

For Safaricom, AI is on both sides of the customer-protection challenge. The company can use automated systems to identify unusual behaviour and suspicious transactions, while fraudsters can use generative AI to create more persuasive impersonations and misleading content. Technology can strengthen detection, but customers still need reliable ways to verify what they see and hear.

Information and Communications and Digital Economy Cabinet Secretary William Kabogo also warned at the summit that deepfakes and AI-generated content posed risks beyond financial fraud, including misinformation and the creation of non-consensual images. The concerns underline how AI-enabled deception can affect individuals, businesses and public trust, not just banking and telecommunications systems.

Safaricom’s customer education efforts are being advanced through its #JiulizeKamaNiUkweli campaign, which encourages customers to question suspicious offers and verify information before acting. The message is particularly relevant when an advertisement or video appears to come from Safaricom or promotes products such as Fuliza.

The campaign’s approach can be distilled into three steps: stop, think and verify. Customers should stop when a message or caller pressures them to act immediately, think critically about offers that promise unusually high returns or extraordinary financial benefits, and verify the information through official Safaricom channels before taking action.

Customers should be cautious of advertisements promising to double Fuliza limits, unsolicited investment opportunities and requests to disclose confidential information. They should never share their M-PESA PIN, passwords or one-time verification codes, or follow unfamiliar links and USSD instructions simply because a message appears to carry Safaricom branding.

Where an offer is suspicious, customers can check Safaricom’s official communication channels or contact customer care for confirmation. Safaricom customers can call 100 from a Safaricom line, while customers using other networks can call +254 722 002 100. For M-PESA services, customers can also use the official *334# menu rather than following instructions from an unverified message.

The broader challenge is that AI-generated content can look and sound authentic, making visual appearance alone an unreliable way to establish whether a message is genuine. Customers need to check the source, assess the claim and independently confirm the offer before sharing information, clicking a link or transferring money.

As AI becomes more deeply integrated into digital transactions, Safaricom’s ability to detect suspicious activity will remain important, but so will cooperation among technology companies, regulators and customers. The dismissal of employees over fraud also highlights the need for internal accountability alongside investment in automated security systems.

The race between AI-enabled fraud and AI-powered protection is likely to intensify as the technology becomes more accessible. For customers, the immediate lesson is straightforward: a convincing video, familiar logo or promise of easy money is not proof that an offer is genuine. Verification remains an essential part of staying safe online.