Cullen O’Keefe, et al., Law-Following AI: Designing AI Agents to Obey Human Laws 94 Fordham L. Rev. 57 (2025).

It is premised on the idea that what AI is doing is not just automatically generating a response but emulating human reasoning, and as such when it develops sufficient reasoning capacity to replace people, it should be imbued with a system of values that it wants to follow. Like a human, it should be “able” to break the law in some circumstances, but it should be programmed to use a set of core values and principles in its reasoning.

It even talks about the obvious solutions that are already established (hard coding it not to break the law) and just dismisses them as outdated. Because obviously if it SOMETIMES outputs info that LOOKS like it resulted from reasoning, it must be possible for a computer to reason like a human. I hate it, I hate it, I hate it.