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and not peer-reviewed.
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 Everything is exploratory and not peer-reviewed.