The most dangerous fraudster in the world today may not be a person at all.
Interpol’s second edition Global Financial Fraud Threat Assessment, released during its annual fraud summit, draws a line in the sand that security practitioners and financial crime investigators cannot afford to ignore. AI-enhanced fraud schemes now generate 4.5 times more revenue than conventional fraud tactics, it said.
Global financial fraud losses reached an estimated $442 billion in 2025
And a new class of autonomous AI, known as agentic AI, is planning and executing an entire fraud campaign from start to finish without any human intervention.
Agentic AI refers to artificial intelligence systems capable of setting goals, making decisions, and taking sequential actions independently. In the context of financial fraud, Interpol documents agentic AI conducting victim reconnaissance, harvesting credentials, infiltrating systems, identifying high-value data, calculating optimal ransom amounts through financial analysis, and generating psychologically tailored ransom notes — all autonomously.
Global financial fraud losses reached an estimated $442 billion in 2025 alone, according to the Global Anti-Scam Alliance data cited in the report. The United States saw fraud losses climb from nearly $4 billion in 2020 to $16.6 billion in 2024.
Also read: U.S. Fraud Losses Soared 25% in 2024 to $12.5 Billion
The 4.5x profitability multiplier that AI delivers means criminal networks now face a straightforward return-on-investment calculation that favors abandoning slower, human-dependent methods entirely.
Dark web marketplaces already offer the infrastructure to support this transition at scale. Interpol identifies the proliferation of Deepfake-as-a-Service platforms selling synthetic identity kits. These are complete with AI-generated video avatars, voice clones, and biometric data that allow attackers to impersonate individuals with just 10 seconds of harvested audio from public sources such as social media. These kits bypass authentication systems that enterprises built their access controls around.
Interpol documented criminal peer-to-peer marketplaces offering end-to-end fraud infrastructure that includes phishing toolkits, fake trading platforms, AI-powered chatbots for victim grooming, encrypted communications and integrated money laundering services.
The investigations also found Eastern European networks selling phishing kits to scam centers in Southeast Asia, which then route proceeds through laundering networks in South Asia. The supply chain for fraud now rivals the operational sophistication of legitimate technology businesses.
Fraud-as-a-Service — where criminal groups offer packaged fraud capabilities to less technically proficient actors for a fee — has further lowered the barrier to entry. Interpol assesses these criminal networks as highly organized, skilled, adaptable, and polycriminal, meaning they operate across multiple crime types simultaneously, including human trafficking and terrorism financing.
The human trafficking dimension deserves particular attention from security operations teams. Hundreds of thousands of victims from nearly 80 nationalities now face trafficking into scam centers — a 54% increase in fraud-related Interpol Notices and Diffusions between 2024 and 2025 shows the pace at which this industrialization advances.
Trafficked individuals inside these compounds increasingly deploy AI-generated imagery to execute sextortion schemes, with instructions to pivot from investment fraud scripts to blackmail if initial approaches fail. Sextortion is no longer an opportunistic afterthought — it is a systematically scripted fallback baked into the fraud playbook.
BEC losses in North America run in the billions annually.
Business Email Compromise remains the most frequently reported fraud type to Interpol globally, with AI now enabling deepfake audio impersonation of CEOs and CFOs during live phone calls to authorize fraudulent wire transfers. BEC losses in North America run in the billions annually.
INTERPOL rates the overall global financial fraud risk as “High”, with projections of significant escalation over the next three to five years across economic, security, and social domains. The agency calls for immediate criminalization of malicious generative AI use for impersonation and voice cloning at scale, enhanced regulatory oversight of virtual asset service providers, and coordinated real-time asset-freezing mechanisms across jurisdictions.
