[@Roach]

It does have to handle the errors in some way or another (including not processing the tags at all), so while I can remove the auto-correct features, it needs to do something with the broken BBCode. Does it scan for the correct closing tag and then return everything in between as plaintext (can also break styling), does it try to unwind until it's at the base level of the post, send that as plaintext, and then parse the rest as BBCode (i.e. the existing system)? Does it send the entire thing as plaintext?

It's a difficult problem to solve.