{
  "annotation_id": "4731039b-5775-493a-a6b9-d3e94d542c37",
  "date_created": "2026-02-17T05:53:20.514000Z",
  "date_modified": "2026-02-17T05:53:20.514000Z",
  "file_hash": "2571aacd20aeda4f0e5d0cd667aec3abc7a9e5ce6956161654efd27d095e04e0",
  "private": false,
  "record": {
    "abstract": "We present an experimental evaluation that assesses the robustness of four open source LLMs claiming function-calling capabilities against three different attacks, and we measure the effectiveness of eight different defences. Our results show how these models are not safe by default, and how the defences are not yet employable in real-world scenarios.",
    "arxiv_id": "2601.09292",
    "authors": [
      "Greta Dolcetti",
      "Giulio Zizzo",
      "Sergio Maffeis"
    ],
    "categories": [
      "cs.CR",
      "cs.AI"
    ],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "title": "Blue Teaming Function-Calling Agents",
    "url": "https://arxiv.org/abs/2601.09292",
    "version": "v1"
  },
  "schema_id": "dorsal/arxiv",
  "source": {
    "execution_id": "4bad78a9-4d6a-45e5-8bcb-5e5eb941d9fb",
    "id": "arXiv Dataset",
    "type": "Model",
    "variant": "snapshot-2026-01-17",
    "version": "0.1.0"
  },
  "user_id": 1000002
}