Couldn't remember a story reference, so collecting a few of these to make finding them again easier.
- Manna, by Marshall Brain: McDonald's micro-management for everyone.
- Reflections on Trusting Trust, by Ken Thompson: Not AI, but why you can't trust anything a computer tells you.
- Coding Machines, by Lawrence Kesteloot: Only slightly less malevolent than Thompson.
- Rounding Corrections, by Sandra Haynes: Sorry for sending you to Jizzmodo, but I don't know of any better link; enable all ad blockers and private browsing before going here. The story's short, not entirely/at all plausible, but one of the few "nice" AIs.
- Avogadro Corporation, by William Hertling: Email classification and auto-responders have no mercy. Or maybe that's just Google. The sequels get kinda cheesy, I prefer the non-anthropomorphized AI in this first one.
- When HARLIE Was One: A classic of a wiseass and often wise computer, and the stupidity of trying to shut it down.