{
  "id": 10994,
  "url": "https://arxiv.org/abs/2607.15207",
  "title": "BadWAM: When World-Action Models Dream Right but Act Wrong",
  "summary": "World-action models (WAMs) are emerging as a promising foundation for embodied control: rather than predicting actions alone, they learn representations that couple action generation with future world prediction. This coupling is often viewed as a source of robustness, interpretability, and safety, as a robot's action can in principle be checked against its imagined future. In this paper, we show that this assumption is fragile. We introduce BadWAM, a unified framework for modeling and evaluatin",
  "authors": "Qi Li, Xingyi Yang, Xinchao Wang",
  "category": "research",
  "topics": "safety-alignment,agents-autonomy",
  "orgs": null,
  "regions": null,
  "published_at": "2026-07-15T20:00:00.000Z",
  "fetched_at": "2026-07-17T05:10:53.887Z",
  "source_slug": "hf-daily",
  "source_name": "HuggingFace Daily Papers",
  "source_homepage": "https://huggingface.co/papers",
  "ethics_ai_record_url": "https://ethics.ai/record/10994",
  "original_url": "https://arxiv.org/abs/2607.15207",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}