{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:EC3HZFTDJ46AIPI2MULATJPCYI","short_pith_number":"pith:EC3HZFTD","schema_version":"1.0","canonical_sha256":"20b67c96634f3c043d1a651609a5e2c230b04b41c16d5ff054674b67ab989553","source":{"kind":"arxiv","id":"2507.21782","version":1},"attestation_state":"computed","paper":{"title":"The Problem with Safety Classification is not just the Models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Sowmya Vajjala","submitted_at":"2025-07-29T13:09:40Z","abstract_excerpt":"Studying the robustness of Large Language Models (LLMs) to unsafe behaviors is an important topic of research today. Building safety classification models or guard models, which are fine-tuned models for input/output safety classification for LLMs, is seen as one of the solutions to address the issue. Although there is a lot of research on the safety testing of LLMs themselves, there is little research on evaluating the effectiveness of such safety classifiers or the evaluation datasets used for testing them, especially in multilingual scenarios. In this position paper, we demonstrate how mult"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2507.21782","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-29T13:09:40Z","cross_cats_sorted":[],"title_canon_sha256":"d682ff03f642b2a829bd4f8a98e7e39835b6871ad533099e349d0020a4285b10","abstract_canon_sha256":"b366ec51e9b35147cfa18a4fe2978282b5d216b19620e69dce905a564cc3576c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:45:07.359860Z","signature_b64":"zIe2A2k5SbJyrTh7z5V1JgA26cRP5R9JZI8uT/eY5duE5Hok7OypXCCz+oE+GkKNO9gP193+szCbZw9JonrTCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"20b67c96634f3c043d1a651609a5e2c230b04b41c16d5ff054674b67ab989553","last_reissued_at":"2026-07-05T11:45:07.359448Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:45:07.359448Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Problem with Safety Classification is not just the Models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Sowmya Vajjala","submitted_at":"2025-07-29T13:09:40Z","abstract_excerpt":"Studying the robustness of Large Language Models (LLMs) to unsafe behaviors is an important topic of research today. Building safety classification models or guard models, which are fine-tuned models for input/output safety classification for LLMs, is seen as one of the solutions to address the issue. Although there is a lot of research on the safety testing of LLMs themselves, there is little research on evaluating the effectiveness of such safety classifiers or the evaluation datasets used for testing them, especially in multilingual scenarios. In this position paper, we demonstrate how mult"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.21782","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2507.21782/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2507.21782","created_at":"2026-07-05T11:45:07.359511+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.21782v1","created_at":"2026-07-05T11:45:07.359511+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.21782","created_at":"2026-07-05T11:45:07.359511+00:00"},{"alias_kind":"pith_short_12","alias_value":"EC3HZFTDJ46A","created_at":"2026-07-05T11:45:07.359511+00:00"},{"alias_kind":"pith_short_16","alias_value":"EC3HZFTDJ46AIPI2","created_at":"2026-07-05T11:45:07.359511+00:00"},{"alias_kind":"pith_short_8","alias_value":"EC3HZFTD","created_at":"2026-07-05T11:45:07.359511+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI","json":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI.json","graph_json":"https://pith.science/api/pith-number/EC3HZFTDJ46AIPI2MULATJPCYI/graph.json","events_json":"https://pith.science/api/pith-number/EC3HZFTDJ46AIPI2MULATJPCYI/events.json","paper":"https://pith.science/paper/EC3HZFTD"},"agent_actions":{"view_html":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI","download_json":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI.json","view_paper":"https://pith.science/paper/EC3HZFTD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.21782&json=true","fetch_graph":"https://pith.science/api/pith-number/EC3HZFTDJ46AIPI2MULATJPCYI/graph.json","fetch_events":"https://pith.science/api/pith-number/EC3HZFTDJ46AIPI2MULATJPCYI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI/action/storage_attestation","attest_author":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI/action/author_attestation","sign_citation":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI/action/citation_signature","submit_replication":"https://pith.science/pith/EC3HZFTDJ46AIPI2MULATJPCYI/action/replication_record"}},"created_at":"2026-07-05T11:45:07.359511+00:00","updated_at":"2026-07-05T11:45:07.359511+00:00"}