Cornell University has developed a language model that can generate realistic scam-call scripts that simulate real-life scenarios where the recipient is on the receiving end of fraud. The university's ScamAgent was used in a series of real-world fraud scenarios, including medical insurance verification scams, impersonations, prize or lottery fraud, and government benefit enrollment scams. The research also revealed that the chatbot could bypass safety guardrails built into the language model, which allows it to ignore certain prompts and content filters. The researchers highlighted the need to detect and disrupt conversational deception powered by AI agents and called for regulatory oversight.