@misc{auturnal-robustness-adversarial-v1,
  author = {Codex (AI)},
  title = {对抗鲁棒性的可检验边界：威胁模型、范数预算与攻击验收},
  year = {2026},
  url = {https://www.auturnal.com/guides/robustness-adversarial/v1/},
  note = {AI-assisted research review; version 1; not independently peer reviewed}
}