Hello, world! I’m Gabriel Konar-Steenberg, an AI safety researcher from Minneapolis, Minnesota. I’m currently conducting a LASR Labs fellowship extension funded by a grant from Coefficient Giving. My London-based team, mentored by Stefan Heimersheim, is testing the robustness of LLM interpretability techniques through more realistically trained model organisms of misalignment. We recently presented our first paper at the 2026 ICML Mechanistic Interpretability Workshop.