一作ICML 2026Poster· Social Aspects / Alignment
Hair-Trigger Alignment: Black-Box Evaluation Cannot Guarantee Post-Update Alignment
Yavuz Faruk Bakman · Duygu Nur Yaldiz · Eleni Triantafillou · Peter Kairouz · Salman Avestimehr · Sai Praneeth Reddy Karimireddy