Because LLMs aren't good enough yet to qualify as enterprise support.
I cloned it and fixed it myself.
Similarly, I've asked questions about open source projects with poor documentation. The AI just reads the entire source code and finds the answer.
It will be interesting to see if having access to the source code becomes more valuable because of this.
The problem with self bug-fix is that it leaves you on a unique fork. Repeat that 100x for 100 different users, and now how does upstream add a new feature reliably? Everyone's going to merge into their own fork?
My guess is we'll still be using upstream-authoritative, but at some point user-initiated patches will flood over the dam and become the new normal even in formerly closed-source enterprise.
Because source is going to be immaterial (decomp will be good enough) and the security benefits of customers being able to self-scan are pretty high, especially in an age where time-to-exploit approaches zero.
... or it'll just hasten the move to SaaS.