← back to paper

Review history

arxiv: 2508.20333 · 2 revisions

Poison Once, Refuse Forever: Weaponizing Alignment for Injecting Bias in LLMs

1. 2026-08-15 CONDITIONAL MODERATE v1.4.0-alltime-deepseek-medium novelty 5.0
   146607 ms 27338 in 16687 out 2026-08-15T16:46:52.319556+00:00
2. 2026-08-05 CONDITIONAL MODERATE v1.4.0-alltime-deepseek-medium novelty 6.0
   130447 ms 27342 in 12616 out 2026-08-05T15:05:20.258630+00:00