@article{whencontextflipssafetybreaksdiagnosingbr, title = {When Context Flips, Safety Breaks: Diagnosing Brittle Safety in Aligned Language Models}, author = {Dasol Choi and Alex Kwon}, year = {2026}, eprint = {2605.27851}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2605.27851}, }