This research introduces Bot Wars, an innovative framework leveraging LLMs to counter phone scams by simulating authentic victim personas that strategically waste scammers' time and resources.

• Creates a two-layer prompt architecture enabling demographically authentic victim personas while maintaining strategic coherence
• Develops formal foundations for strategy emergence through chain-of-thought reasoning without explicit optimization
• Demonstrates effectiveness using a dataset of 3,200 scam dialogues
• Establishes a novel approach to cybersecurity defense by turning AI against scammers

This research matters for security professionals as it offers a scalable, automated solution to combat the growing phone scam epidemic, potentially reducing successful fraud attempts while increasing costs for scammers.

Bot Wars Evolved: Orchestrating Competing LLMs in a Counterstrike Against Phone Scams