From my understanding of cyberpunk, AI isn't really an essential ingredient. Cyberpunk feels very human.
AIs are either of the "weak" kind, like a robot guard being able to recognize and attack intruders but not much more, or the indistinguishable-from-human kind of Blade Runner replicants. The intelligent but definitely not human kind like HAL from 2001. The kind I tend to think of when talking about AI-based stories don't seem to be that common in the cyberpunk genre.
We do see a lot of things like mind uploading, brain augmentation, people losing their humanity to technology, etc... but it is always centered on the human mind.
I think hostile AI isn't really an essential ingredient in cyberpunk for me, but some sort of AI, and often coupled with some sort of conflict between whether the AI is hostile or benevolent. I associate cyberpunk very much with the grey area between man and machine.
I wouldn't mind a Nosedive redux episode where you discover she is part of an experiment by an AI to see how quickly people can drop off and keep them bumping along the bottom of society. Sort of social Flappy Bird.
Well...Wintermute wasn't essentially hostile to humanity - it just had it's own goals. I find that way more cyberpunk than a dystopian AI bent on humanity's destruction.
- hostile AI
- Corporations and Government surveilance and going dark
- DNA / implant chips / biometric border controls
- Borders protected by armed drones / robots
- Underground Identities markets