Z. Wang, “STAR-1: Safer Alignment of Reasoning LLMs with 1K Data”, AAAI, vol. 40, no. 44, pp. 37988–37997, Mar. 2026.