The Global Fraud & Crime Threat Landscape

The global fraud crisis

Global fraud losses exceeded $442 billion in 2025 — a figure that continues to accelerate as AI tools lower the cost of executing sophisticated fraud campaigns and raise their profitability by 4.5×. What once required organised criminal teams with specialist skills can now be deployed by a single operator with access to commercially available AI tools.

Industrial fraud centres — operating legally as call centres or technology companies — now exist in more than 72 countries. These are not opportunistic criminal enterprises. They are structured businesses with management hierarchies, KPIs, shift patterns, and specialised training programmes. They target banks, telecoms, insurance companies, and individuals with coordinated, AI-assisted attack campaigns.

Regional threat breakdown

Africa under siege

South Africa sits at the intersection of two converging threat streams: sophisticated financial fraud targeting the continent's most developed banking sector, and violent polycriminality — kidnapping, hijacking, armed robbery — that has become increasingly organised and networked.

The Safer City Solutions intelligence network tracks 26,426 incidents across 5,483 South African suburbs continuously. The data reveals a consistent pattern: crime does not operate in isolation. Hijacking networks overlap with kidnapping operations. Commercial crime syndicates use proceeds to fund armed robbery infrastructure. The same individuals appear across multiple crime categories — the definition of polycriminality.

Traditional law enforcement and corporate security tools treat these as separate problems. Safer City Solutions was built on the premise that they are one problem — and that AI capable of correlating across categories is the only response capable of matching the threat.

The deepfake fraud acceleration

Deepfake audio and video technology has removed one of the last friction points in high-value fraud: voice verification. Fraudsters can now generate real-time audio impersonation of any target — a company CEO, a bank's fraud verification agent, a family member — with less than 30 seconds of source audio.

The consequence for financial institutions is severe. Authorised push payment fraud — where victims are convinced by a trusted voice to transfer funds — cannot be defeated by voice-based verification alone. Device-level fraud signals (IMEI status, SIM-swap history, network anomalies) become a critical secondary verification layer.

The kidnapping & trafficking dimension

The Safer City intelligence network tracks 419 active kidnapping cases in South Africa. Of these, 58% are express kidnappings — short-duration, high-frequency incidents targeting high-value individuals in predictable locations. A further 28% are kidnap-for-ransom operations with corporate or family targets.

What makes the kidnapping threat increasingly dangerous is its integration with other crime categories. The same networks that operate hijacking syndicates are now running express kidnapping operations — using vehicle intelligence gathered during hijacking operations to identify targets, assess wealth signals, and plan extraction routes.

Why traditional security fails

• Reactive, not predictive — security operations respond to incidents that have already occurred
• Siloed by crime category — fraud teams, physical security teams, and law enforcement share no data
• Human-speed analysis — intelligence cycles measured in days cannot match AI-speed threat execution
• No cross-network visibility — a device flagged by one bank's fraud system is invisible to every other bank
• Retrospective reporting — incident reports describe what happened, not where it will happen next
• No polycriminality correlation — the same individual operating across fraud, kidnapping and hijacking appears as three separate suspects in three separate systems

The Safer City Solutions response