Original title: Effective Interventions Against AI-Enhanced Scams
Kyle Fredrickson · September 2026
Scams cause massive financial damage worldwide, and criminals are now using artificial intelligence to alter the economics and operational limits of running these schemes. The author studied which defenses can still work against these new AI-driven methods. To do this, the researcher built a simple mathematical model of scam profits to see how interventions change operations over time. The study found three key factors: how often people report scams, having a centralized reporting system, and the accuracy of those reports. These three elements multiply together to lower the expected victims and revenue per scam channelA specific method or platform scammers use to reach victims, like a fake email campaign or a fraudulent website. while driving up criminals' costs. Targeting all three areas, or achieving even modest reporting rates against high-value scam infrastructure, can significantly hurt scam profitability.
You and your neighbors are trying to stop package thefts from your building. If only a few people report suspicious cars, and those reports go to different phone numbers, and some descriptions are wrong, the police can't catch anyone. In this paper, the same three factors—how many victims report, whether reports go to one central database, and how accurate they are—multiply to determine how much profit scammers can make.
Scams cost the world over $440 billion in 2025, and AI is making them worse. This paper shows that three simple reporting improvements—rate, centralization, and accuracy—multiply to slash scam profits. Even modest reporting against key infrastructure could reverse the trend.
The abstract does not provide empirical validation of the model or test specific interventions. It focuses on asymptotic effects and does not quantify exact thresholds for reporting rates. The analysis is theoretical and may not capture all real-world complexities of scam operations.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Effective Interventions Against AI-Enhanced Scams — Kyle Fredrickson
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.