[xg15]

Sounds like a perfect use case for text generators to me. If you receive a mail that you're 100% sure is a scam or phishing attempt, you could pass it to ChatGPT and have it teergrube the scammer into some endless conversation that binds as many resources on the scamner's side as possible.

(emphasis on "100% sure" though. If such a system was widely deployed, it could also quickly turn into a Kafkaesque horror show if legitimate messages get caught in it)