Here's an older submission about using ChatGPT to de-obfuscate the more basic methods:

https://news.ycombinator.com/item?id=38150096

Some comments claim that it can break some of the more complex techniques presented in the article. I've tried it a few times myself with varying results that tend towards not working.