Hey, I'm Jasmine!

I research technical safeguards for artificial intelligence, and study CS and English at Cornell as a Milstein and Telluride Scholar.

I care about making the development of advanced AI go well. I'm currently a MATS scholar under Victoria Krakovna, developing honeypots for scheming propensity. Most recently, I developed agent security evaluations at Gray Swan AI and developed methods for LLM honesty at the Center for AI Safety. I also founded and lead Cornell AI Alignment! See my research and what I'm up to now.

Other things I love: persimmons, progress and flourishing, optimistic science fiction and beautiful poetry, AI for science, the 10% pledge 🔸, outdoor adventures, tomato egg noodles, self-integrity, and learning what, and how, others love. Please reach out and say hi (jasminexinzeli@gmail.com)!