{
  "id": 5606,
  "url": "https://arxiv.org/abs/2605.15204v1",
  "title": "SDOF: Taming the Alignment Tax in Multi-Agent Orchestration with State-Constrained Dispatch",
  "summary": "Multi-agent orchestration frameworks such as LangChain, LangGraph, and CrewAI route tasks through graph-based pipelines but do not enforce the stage constraints that govern real business processes. We present SDOF, a framework that treats multi-agent execution as a constrained state machine. SDOF operates through two primary defensive layers, implemented by three components: (1) an Online-RLHF Specialized Intent Router trained via Generative Reward Modeling (GRPO) and (2) a StateAwareDispatcher ",
  "authors": "Zhantao Wang",
  "category": "research",
  "topics": "safety-alignment,agents-autonomy",
  "orgs": null,
  "regions": null,
  "published_at": "2026-04-20T12:51:39.000Z",
  "fetched_at": "2026-07-14T16:31:53.165Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/5606",
  "original_url": "https://arxiv.org/abs/2605.15204v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}