A new INTERPOL report has produced a headline that is already travelling on its own: artificial intelligence is now behind more than half of cybercrime in Africa. The figure is real and it comes from a credible source. It is also worth reading slowly, because what the report actually measured is narrower than the headline suggests — and the part that should worry you is not the percentage.

What the report says

INTERPOL's African Cyberthreat Assessment Report 2026, published on 3 August, states that "artificial intelligence is enabling 55 per cent of reported cybercrimes across Africa." The 40-page report draws on survey data from 36 African member countries, and its through-line is that cybercrime on the continent has shifted from scattered incidents into an industrialised, borderless business.

The hard numbers underneath are the story:

• Losses have more than doubled since 2024 — from USD 192 million to USD 484 million.
• 72 per cent of surveyed countries reported the presence of scam centres, concentrated in Southern and West Africa.
• Business email compromise — where a criminal poses as a trusted contact to redirect a payment — has risen sharply, with AI used to generate convincing correspondence.
• One technology partner, TrendAI, recorded roughly 600,000 sextortion detections, a category where deepfakes and AI-generated imagery are increasingly in play.

What the 55% figure is, and isn't

Read the sentence precisely: AI is enabling 55 per cent of reported cybercrimes, according to a survey of member countries. Three qualifiers do a lot of work there.

Reported means this counts the crimes that reached the authorities — a fraction of the real total, and a fraction that varies enormously between countries. Survey means the figure reflects what 36 national agencies told INTERPOL they are seeing, not a forensic audit of seized infrastructure; it is a considered assessment, not a measurement, and the report offers no confidence interval around it. And enabling is a broad tent: it stretches from a scam email drafted by a chatbot to a fully synthetic deepfake video, lumping a small assist and a wholesale fabrication into the same bucket.

So "55 per cent" is best read as what law-enforcement agencies across Africa believe about how often AI now touches the crimes they see — which is a useful signal of how fast these tools have spread into criminal workflows. It is not a claim that AI committed half of Africa's cybercrime, and it should not be quoted as one.

Why the trend is not in doubt

None of that caution applies to the direction of travel, which is well evidenced and consistent with what security teams are reporting elsewhere. The reason AI shows up so often is simple economics: it removes the two things that used to limit a scammer. It erases the language barrier — a business email compromise attempt no longer arrives in broken English — and it removes the labour ceiling, letting one operator run the volume that used to take a team. That is what "industrialised" means in practice, and it is why losses can double in two years while the number of people running the scams does not.

The most human cost sits in that sextortion figure. Cheap, convincing, AI-generated imagery lowers the barrier to a crime that targets individuals directly, and 600,000 detections from a single partner is a floor, not a ceiling.

The takeaway

Treat the 55 per cent the way you would any survey headline: as a real signal wrapped in soft measurement. The precise number will not survive close contact with better data, and quoting it as gospel does the report a disservice. What will survive is the shape INTERPOL is describing — cybercrime becoming faster, cheaper to run, harder to detect, and less dependent on the criminal's own skill — because that is the predictable result of putting capable, cheap AI tools within everyone's reach, including the people you would least want to have them.