ChatGPT already recognises common scam emails, so it probably won't be that cooperative. Interactions with scammers aren't that helpful for training LLMs, and I

ChatGPT already recognises common scam emails, so it probably won't be that cooperative. Interactions with scammers aren't that helpful for training LLMs, and I would expect accounts whose conversations consist solely of interactions associated with scams will likely end up being flagged up and banned. In theory you could solve this with a custom LLM but.... that's a lot of time, compute costs and effort to waste scammer time which is almost worthless