{
  "id": "over-refusal",
  "code": "PTL-0176",
  "term": "Over-Refusal",
  "aliases": [
    "exaggerated safety",
    "false refusal",
    "overrefusal"
  ],
  "category": "failure-modes",
  "definition": "Over-refusal is a model declining safe requests because they superficially resemble harmful ones, such as refusing to explain how to kill a Python process.",
  "description": "It is the cost side of safety training, since overly cautious models frustrate users and withhold legitimate help. Test suites such as XSTest pair safe prompts with unsafe look-alikes to measure the balance, and methods such as deliberative alignment aim to reduce refusals while improving robustness.",
  "example": null,
  "broader": [],
  "narrower": [],
  "related": [
    "guardrails",
    "rlhf",
    "deliberative-alignment",
    "constitutional-classifiers"
  ],
  "introduced": 2023,
  "added": "2026-10-10",
  "sources": [
    {
      "title": "XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models",
      "authors": "Röttger et al.",
      "year": 2023,
      "url": "https://arxiv.org/abs/2308.01263"
    }
  ],
  "url": "https://protologue.com/t/over-refusal/",
  "citation": "Protologue. (2026). Over-Refusal. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0176). https://protologue.com/t/over-refusal/"
}