Effective Interventions Against AI-Enhanced Scams

Global direct losses from scams were estimated at $442 billion in 2025, with reported losses in the United States increasing by nearly 400% between 2020 and 2025. This research investigates effective interventions against AI-enhanced scams, which alter the economics and operational bottlenecks of traditional scamming. A model of scam profits identifies that increasing reporting rates, centralizing reporting mechanisms, and improving report accuracy are crucial, as these factors multiply in their effect on reducing revenue per scam channel.

Why it matters

The rapid increase in scam losses, exacerbated by AI, presents a significant and evolving threat to economic stability and public trust. Understanding the new operational dynamics of AI-driven scams is critical for developing robust and effective countermeasures that can scale with the threat.

Key insights

• Global scam losses reached an estimated $442 billion in 2025.
• US reported scam losses escalated by nearly 400% from 2020 to 2025.
• The emergence of AI in scamming fundamentally changes scam economics and operational bottlenecks.
• A model of scam profits suggests that reporting rate, centralization of reporting, and report accuracy are key levers.
• These three levers have a multiplicative effect on reducing the expected number of victims and revenue per scam channel.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2609.00806