{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:3XGKVZ7N2X3QZMTXCD3SYGUBWT","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"3727b343e63e752f64d891f0f76f9a624ea0c73451bcd3f610ff51e4555ab629","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-25T15:51:15Z","title_canon_sha256":"3270cb29772400247e89960336f93fb593135f6540b8948c998d8c9ddcc03490"},"schema_version":"1.0","source":{"id":"2308.13449","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2308.13449","created_at":"2026-07-05T06:44:44Z"},{"alias_kind":"arxiv_version","alias_value":"2308.13449v1","created_at":"2026-07-05T06:44:44Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.13449","created_at":"2026-07-05T06:44:44Z"},{"alias_kind":"pith_short_12","alias_value":"3XGKVZ7N2X3Q","created_at":"2026-07-05T06:44:44Z"},{"alias_kind":"pith_short_16","alias_value":"3XGKVZ7N2X3QZMTX","created_at":"2026-07-05T06:44:44Z"},{"alias_kind":"pith_short_8","alias_value":"3XGKVZ7N","created_at":"2026-07-05T06:44:44Z"}],"graph_snapshots":[{"event_id":"sha256:455f0feaea005312eb0cea8a1f07773ea457ef3d4d21a3408a55363221888b51","target":"graph","created_at":"2026-07-05T06:44:44Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2308.13449/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"From the perspective of content safety issues, alignment has shown to limit large language models' (LLMs) harmful content generation. This intentional method of reinforcing models to not respond to certain user inputs seem to be present in many modern open-source instruction tuning datasets such as OpenAssistant or Guanaco. We introduce a novel insight to an instruction-tuned model's performance affected by the presence of alignment in supervised fine-tuning dataset. To be specific, we noticed that alignment acts as if it is poisoning the instruction dataset. Experimentally, we demonstrate tha","authors_text":"Aibek Bekbayev, James Yamazaki, Sungbae Chun, Yerzat Dulat","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-25T15:51:15Z","title":"The Poison of Alignment"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.13449","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:cd66a8590e50cfc40f15677717e0d5860cbf43199bedb95c47591e98a4feb297","target":"record","created_at":"2026-07-05T06:44:44Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"3727b343e63e752f64d891f0f76f9a624ea0c73451bcd3f610ff51e4555ab629","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-25T15:51:15Z","title_canon_sha256":"3270cb29772400247e89960336f93fb593135f6540b8948c998d8c9ddcc03490"},"schema_version":"1.0","source":{"id":"2308.13449","kind":"arxiv","version":1}},"canonical_sha256":"ddccaae7edd5f70cb27710f72c1a81b4e7eb902bfaaf20318c5c3accfc78b504","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ddccaae7edd5f70cb27710f72c1a81b4e7eb902bfaaf20318c5c3accfc78b504","first_computed_at":"2026-07-05T06:44:44.153801Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T06:44:44.153801Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"THeXMiaR3Pkc7MvpKUcPwisA6WVNjKXGOoEUVinWYJ4nt1PuNWafkL9iKsmwgE7XaDJSsQ43x5MckeQZiLj3DA==","signature_status":"signed_v1","signed_at":"2026-07-05T06:44:44.154376Z","signed_message":"canonical_sha256_bytes"},"source_id":"2308.13449","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:cd66a8590e50cfc40f15677717e0d5860cbf43199bedb95c47591e98a4feb297","sha256:455f0feaea005312eb0cea8a1f07773ea457ef3d4d21a3408a55363221888b51"],"state_sha256":"c8b429d4a040471c8319ad18f18f78c02d58df72c3c0dbe5e79bebc96646b1b3"}