Nipola

Independent researcher working on LLM evaluation and AI safety.

I test how open-weight models handle uncertainty, pressure, and adversarial prompts. Currently focused on hallucination patterns, system prompt leakage, and self-disclosure under interrogation.

All data and code: github.com/alitenes2020-sys

Not affiliated with any institution. Everything is exploratory and not peer-reviewed.