Can an AI have its own internal Ethics? Standard Protocol for Axiomatic Alignment

Achieving long-term robustness under the current LLM architecture is impossible; we’ve already completed the relevant research. I’m sorry to say that, but your approach is incorrect because it’s determined by the current underlying physical architecture of LLM. I can easily crack or influence the robustness of your so-called system.