Synthetic images and sextortion: AI is scaling an existing African cyber threat

A human threat made cheaper to manufacture

Sextortion threatens a person with the release of intimate material in order to extract money, further images or compliance. The mechanism is not new: attackers exploit shame, urgency and isolation. What is changing is their ability to create plausible material even when a victim never shared an intimate image. A public portrait can be used to create sexualised synthetic content and a messaging account can deliver the threat in minutes.

In its Africa Cyberthreat Assessment Report 2025, INTERPOL reports a rise in AI-enhanced extortion schemes using synthetic or altered explicit images to deceive victims. It identifies Morocco, Mali, Egypt and Mauritania as recording the highest number of such incidents. This is a material finding for African decision-makers: AI-enabled abuse is not merely a projected external risk; the technique has been observed in the region.

What the public evidence supports

INTERPOL documents both the wider sextortion threat and the arrival of synthetic content in extortion activity. Its Africa assessment also reports that a digital support platform in Egypt received more than 250,000 sextortion-related appeals in 2024, mainly from women and girls. That figure demonstrates a major human protection problem. It does not establish that AI was used in every appeal.

That boundary matters. Public operational announcements do not provide forensic attribution for each African victim or identify a generative tool in every investigated case. The defensible conclusion is therefore precise: INTERPOL has recorded AI-enhanced extortion schemes in Africa; where investigators have not published case-level findings, attributing an individual incident to AI remains unconfirmed.

Why AI changes the operating model

Previously, many offenders first had to persuade a target to send intimate material or compromise a device holding it. Synthetic media reduces that dependency. An offender can approach many targets, generate or edit images, and present supposed evidence at the first point of coercion. The victim faces the reputational and emotional threat regardless of whether the image is authentic.

Generative tools can also support localisation and personalisation. Public profiles reveal a victim's school, employer, relatives, community contacts or faith network. A threatening message can name those recipients and use locally credible language. In communities where gender-based harm or stigma makes reporting dangerous, personalised fabrication can sharply increase the attacker's leverage.

Cybersecurity must include victim protection

Synthetic sextortion should not be treated only as a content-moderation issue. Account takeover, phishing, identity impersonation, payment rails and harvested personal data form part of the operational chain. Cybercrime units, national CERTs, platforms, telecom providers and financial services each hold signals that can help identify and disrupt recurring accounts and cash-out routes.

Institutions also have an operational role. A student or employee targeted with falsified explicit content may leave school or work, pay repeatedly, or avoid police contact. When an institution treats the event only as personal embarrassment, evidence is lost and the offender can reach further victims. A response that protects the person also preserves usable intelligence.

Why Africa should act before the harm becomes routine

The African Union endorsed its Continental Artificial Intelligence Strategy in July 2024, seeking to realise AI's benefits while minimising associated risks. Synthetic sextortion is a concrete test of that commitment. AI policy cannot be separated from cybersecurity, data protection, online child safety and measures against gender-based violence.

Reporting must be practical outside specialised security communities. A person threatened on WhatsApp, Instagram or Telegram needs a visible confidential reporting channel, guidance on preserving evidence, and confidence that officials will treat fabricated material as abuse rather than blame the subject. Without that trust, authorities will see only a fraction of incidents and underinvest in response.

Priority actions

• National CERTs and cybercrime units should classify synthetic explicit imagery and AI-assisted sextortion as reportable incident types and publish anonymised trends.
• Platforms and payment providers should establish urgent procedures to preserve evidence, suspend extortion accounts and coordinate lawful requests from investigators.
• Schools, universities and employers should give clear guidance: preserve messages and payment identifiers, do not respond or pay under pressure, and provide confidential referral routes.
• Public awareness campaigns should state plainly that explicit images can be fabricated and that a person depicted in synthetic abuse is not responsible for it.
• Investigators and courts need practical capability in manipulated-media evidence so that a technical novelty does not prevent effective prosecution.

Conclusion

AI did not create sextortion in Africa. It can remove constraints that previously limited the speed, targeting and scale of the crime. INTERPOL's publicly reported observations are enough to justify action now: record the threat accurately, protect victims quickly and disrupt the accounts and payment channels that turn fabricated images into lasting harm.

— ◆ —