What we know

Africa Business Communities published a report on 19 November 2026, bylined to Nicholas Mulila, that looks at Safaricom's efforts to shield its customers from scams built with artificial intelligence S1. The headline frames it as an inside account of a fight to keep customers safe from AI scams S1.

Beyond that, the material available does not give figures, named executives, timelines or the specific tools involved S1. This article therefore sets out what the report covers and why the subject matters, without speculating on details it does not contain.

What "AI scams" means in plain English

Artificial intelligence, in this context, is software that generates text, images or audio on demand. Fraudsters use it to smooth out the flaws that used to give a scam away: the clumsy grammar in a phishing message, the obviously fake voice on a call, the implausible document.

That matters because many defences, including customers' own instincts, were partly built on those flaws. When a message reads cleanly and a voice sounds familiar, the person receiving it has to rely on something more than a gut check.

Why a telecoms company cares

For an operator that carries both messages and money, fraud is a commercial problem, not only a safety one. A convincing scam that runs across a network does damage twice: to the customer who loses money, and to the willingness of everyone else to trust the channel.

Trust is the asset telcos sell. It underpins airtime, mobile money, and any move into lending, insurance or merchant payments. It also shapes how much a company spends on call centres, refunds, investigations and compliance staff.

Banks, merchants and regulators watch the same signals. A network seen as a soft target for fraud tends to attract louder supervision and harder negotiations with partners.

What the report does not settle

The available excerpt does not say which scams Safaricom has seen, how many cases it has handled, or what measures it has introduced S1. It carries no direct comment from the company beyond the framing of the headline S1. Readers should treat claims about the scale of the problem as unconfirmed until Safaricom sets them out publicly.

What to watch

Whether the company puts numbers behind the effort: cases reported, accounts blocked, refunds paid. Whether it names specific countermeasures, such as warnings inside its app or extra checks on transfers. And whether the issue draws fresh guidance from the regulators who oversee mobile money and communications.

The wider test is whether operators treat AI-enabled fraud as shared infrastructure, with data pooled across the industry, or as a private race each company runs alone.

The bottom line

Artificial intelligence lowers the cost of running a scam and raises the cost of stopping one. For Safaricom, and for the customers deciding whether to send money through its rails, the answer will show up in how openly the problem is measured S1.

Sources
  1. Nicholas Mulila: Inside Safaricom’s fight to keep customers safe from AI scams Africa Business Communities · news.google.com