Cross-Generational Transfer of Adversarial Attacks Reveals Non-Monotonic Safety Alignment in LLMs

Abstract

Safety alignment in LLMs does not improve monotonically across model generations. Studying four generations of Google's Gemma family (7B-31B) with quality-diversity evolution (MAP-Elites) as an automated red-teaming probe, we find that Gemma 3 (12B) exhibits 68.7% +/- 5.7% attack success rate (ASR; mean +/- std, 3 seeds), significantly higher than its predecessor Gemma 2 (45.5% +/- 7.2%; p = 0.030, paired bootstrap) and its successor Gemma 4 (33.9% +/- 1.8%). Replaying evolved attack archives ac

Categories

Framework mappings

Suggested from the entry's categories.

@misc{mitra2026crossgenerational,
  title = {{Cross-Generational Transfer of Adversarial Attacks Reveals Non-Monotonic Safety Alignment in LLMs}},
  author = {Subhadip Mitra},
  year = {2026},
  month = may,
  eprint = {2606.00813},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2606.00813}
}