I have a friend--a software engineer--who was convinced when ChatGPT came out that it would be able to do things like invert sha1 hashes, if only you found the right prompt.
There's a lot of magical thinking around LLMs, but in the end they're "just" fancy Markov chains.
I'm sure it's possible for some SHA1 hashes along the lines of "password1234".
A shockingly high number of SHA1 hashes with their cleartext was available on Google Code back in the day. I assume you can crawl github for the same now.
I don't know why you've been downvoted to oblivion but I DO find this thoroughly uninteresting. I am however interested in why anyone WOULD find it interesting.
