{
  "id": 5224,
  "url": "https://arxiv.org/abs/2604.26516v1",
  "title": "Lyapunov-Guided Self-Alignment: Test-Time Adaptation for Offline Safe Reinforcement Learning",
  "summary": "Offline reinforcement learning (RL) agents often fail when deployed, as the gap between training datasets and real environments leads to unsafe behavior. To address this, we present SAS (Self-Alignment for Safety), a transformer-based framework that enables test-time adaptation in offline safe RL without retraining. In SAS, the main mechanism is self-alignment: at test time, the pretrained agent generates several imagined trajectories and selects those satisfying the Lyapunov condition. These fe",
  "authors": "Seungyub Han, Hyungjin Kim, Jungwoo Lee",
  "category": "research",
  "topics": "safety-alignment,agents-autonomy,environment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-04-29T10:32:18.000Z",
  "fetched_at": "2026-07-14T16:31:35.575Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/5224",
  "original_url": "https://arxiv.org/abs/2604.26516v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}