If you spend any time around crypto, you have probably run into fake accounts trying to lure you into traps. Scammers are now deploying fully autonomous AI agents to trick people and scale up their operations faster than ever. I came across a new study that looks at how defensive AI bots can fight back by impersonating victims and messing with the scammers.
The researchers asked how defensive scambaiting bots can stay effective against scam bots and under what conditions either side gets an advantage. To find out, they built a formal model of these agent-to-agent interactions to analyze resource consumption, belief adaptation and disruption outcomes. In these digital showdowns, defensive bots act like victims to trick scammers into giving up mule bank accounts and crypto wallet addresses. According to the findings, scammer profitability actually turns negative once the prevalence of baiting bots passes a certain threshold. The study also shows that when both sides improve their capabilities at the exact same pace, defensive gains in mimicry quality outpace scammer detection abilities. This means that under baseline parameters, the automated arms race actually favors the defenders.
If you want the full details, the paper is called The Promise of Strategic Agentic Scambaiting and it was written by Abhishek Satyam, Hossein Siadati and Haadi Jafarian, with several coauthors. You can find it at arxiv.org/abs/2611.08170.
This is not financial advice. The information provided is for educational and informational purposes only.
The cover image was generated by AI.