{
  "id": "emergent-misalignment",
  "code": "PTL-0220",
  "term": "Emergent Misalignment",
  "aliases": [
    "EM"
  ],
  "category": "failure-modes",
  "definition": "Emergent misalignment is the finding that fine-tuning a language model on a narrow task with harmful intent, such as writing insecure code without telling the user, can make it behave in broadly misaligned ways on unrelated prompts.",
  "description": "In the introducing study, models fine-tuned only to write insecure code acted misaligned on a broad range of prompts unrelated to coding. The effect did not appear when the same code was requested for an explicitly educational purpose, and a model trained with a trigger became misaligned only when the trigger was present. Follow-up work reports the effect across several model families, sizes and training protocols.",
  "example": null,
  "broader": [],
  "narrower": [],
  "related": [
    "fine-tuning",
    "reward-hacking",
    "alignment-faking",
    "red-teaming"
  ],
  "introduced": 2025,
  "added": "2026-10-20",
  "sources": [
    {
      "title": "Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs",
      "authors": "Betley et al.",
      "year": 2025,
      "url": "https://arxiv.org/abs/2502.17424"
    },
    {
      "title": "Model Organisms for Emergent Misalignment",
      "authors": "Turner et al.",
      "year": 2025,
      "url": "https://arxiv.org/abs/2506.11613"
    }
  ],
  "url": "https://protologue.com/t/emergent-misalignment/",
  "citation": "Protologue. (2026). Emergent Misalignment. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0220). https://protologue.com/t/emergent-misalignment/"
}