I’ll interpret this as a request for a “modern” approach, as much fiction predates the computer age or shows no realistic depiction.
Let me point out this answer from last year, where I conclude with:
My take on it: data acquisition and study is slower than the projected hardware Moore curve. … When a solid plan is ready, the hardware for a full-scale human brain implementation will cost less than a million dollars, but they'll start with smaller systems like mice, dogs, etc. If the hardware is custom, prototypes and small batches will provide hardware for the mice etc. If it can run on the general purpose high-performance computer (by then not ranked as a supercomputer) you know someone's going to try it long before it's ready.
Hardware capable of running AI will exist at a University level before anybody can actually write an AI. It will be available enough that students will “try” things that are not ready for production, possibly without the concent of the school or their professors.
A hack, half-baked AI, may actually work and be kept in secret. It will not have failsafes and isolation like an official project would. It will not be totally sane. It will get out and grow without being under anybody’s control.