{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:3PO433BIFWKXUE2WVONHBR5YKZ","short_pith_number":"pith:3PO433BI","canonical_record":{"source":{"id":"2503.08919","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-03-11T22:04:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1d019fba93699be5a9700ba3dcc870ac55a4e6f984e7517be84a3ac6da39a660","abstract_canon_sha256":"aae3f99d7e455e304af1c70643eae5c4fcd17c5c1c31726bc9e3f3c55d079e62"},"schema_version":"1.0"},"canonical_sha256":"dbddcdec282d957a1356ab9a70c7b8565c14affba9762f1ec12f57ab52b9157b","source":{"kind":"arxiv","id":"2503.08919","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.08919","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"arxiv_version","alias_value":"2503.08919v1","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.08919","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"pith_short_12","alias_value":"3PO433BIFWKX","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"pith_short_16","alias_value":"3PO433BIFWKXUE2W","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"pith_short_8","alias_value":"3PO433BI","created_at":"2026-07-05T10:29:43Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:3PO433BIFWKXUE2WVONHBR5YKZ","target":"record","payload":{"canonical_record":{"source":{"id":"2503.08919","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-03-11T22:04:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1d019fba93699be5a9700ba3dcc870ac55a4e6f984e7517be84a3ac6da39a660","abstract_canon_sha256":"aae3f99d7e455e304af1c70643eae5c4fcd17c5c1c31726bc9e3f3c55d079e62"},"schema_version":"1.0"},"canonical_sha256":"dbddcdec282d957a1356ab9a70c7b8565c14affba9762f1ec12f57ab52b9157b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:29:43.740182Z","signature_b64":"JHOyxFbJXd4WO8/BNYuXGnWjURIFpAOpYTVVNcllNSGi9ZUWh9lwafA/QQRLn02+ue4goYiILJ20665NFDAHBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dbddcdec282d957a1356ab9a70c7b8565c14affba9762f1ec12f57ab52b9157b","last_reissued_at":"2026-07-05T10:29:43.739686Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:29:43.739686Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2503.08919","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:29:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"mmeb/Mfg95ATlb/F1d5WkyCPNUPYNhsCAgbUEE8tR/VAA0xNZXEUmtmtYheZlyryNpQuSE4VBZo6pZU4L6ucBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T11:25:24.485163Z"},"content_sha256":"3a8d64c985c3b88386bb04c981747f76d524de418329fcc9b380b02fd03b5cf4","schema_version":"1.0","event_id":"sha256:3a8d64c985c3b88386bb04c981747f76d524de418329fcc9b380b02fd03b5cf4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:3PO433BIFWKXUE2WVONHBR5YKZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Backtracking for Safety","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bilgehan Sel, Dingcheng Li, Ming Jin, Phillip Wallis, Siddhartha Reddy Jonnalagadda, Vaishakh Keshava","submitted_at":"2025-03-11T22:04:22Z","abstract_excerpt":"Large language models (LLMs) have demonstrated remarkable capabilities across various tasks, but ensuring their safety and alignment with human values remains crucial. Current safety alignment methods, such as supervised fine-tuning and reinforcement learning-based approaches, can exhibit vulnerabilities to adversarial attacks and often result in shallow safety alignment, primarily focusing on preventing harmful content in the initial tokens of the generated output. While methods like resetting can help recover from unsafe generations by discarding previous tokens and restarting the generation"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.08919","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/2503.08919/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:29:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"oEosLuReB//LSITyqArzkaf4W52t24QWcsB0V//fy4Oom1DIrpNvX0T1tMokRDUquPTfcDqXdEF4J9fBskPHCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T11:25:24.485649Z"},"content_sha256":"c35a8639d1958600b4f269c92f15fe977855df8de1dfc2fc35b6b763497bd084","schema_version":"1.0","event_id":"sha256:c35a8639d1958600b4f269c92f15fe977855df8de1dfc2fc35b6b763497bd084"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3PO433BIFWKXUE2WVONHBR5YKZ/bundle.json","state_url":"https://pith.science/pith/3PO433BIFWKXUE2WVONHBR5YKZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3PO433BIFWKXUE2WVONHBR5YKZ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-08T11:25:24Z","links":{"resolver":"https://pith.science/pith/3PO433BIFWKXUE2WVONHBR5YKZ","bundle":"https://pith.science/pith/3PO433BIFWKXUE2WVONHBR5YKZ/bundle.json","state":"https://pith.science/pith/3PO433BIFWKXUE2WVONHBR5YKZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3PO433BIFWKXUE2WVONHBR5YKZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:3PO433BIFWKXUE2WVONHBR5YKZ","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":"aae3f99d7e455e304af1c70643eae5c4fcd17c5c1c31726bc9e3f3c55d079e62","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-03-11T22:04:22Z","title_canon_sha256":"1d019fba93699be5a9700ba3dcc870ac55a4e6f984e7517be84a3ac6da39a660"},"schema_version":"1.0","source":{"id":"2503.08919","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.08919","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"arxiv_version","alias_value":"2503.08919v1","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.08919","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"pith_short_12","alias_value":"3PO433BIFWKX","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"pith_short_16","alias_value":"3PO433BIFWKXUE2W","created_at":"2026-07-05T10:29:43Z"},{"alias_kind":"pith_short_8","alias_value":"3PO433BI","created_at":"2026-07-05T10:29:43Z"}],"graph_snapshots":[{"event_id":"sha256:c35a8639d1958600b4f269c92f15fe977855df8de1dfc2fc35b6b763497bd084","target":"graph","created_at":"2026-07-05T10:29:43Z","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/2503.08919/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) have demonstrated remarkable capabilities across various tasks, but ensuring their safety and alignment with human values remains crucial. Current safety alignment methods, such as supervised fine-tuning and reinforcement learning-based approaches, can exhibit vulnerabilities to adversarial attacks and often result in shallow safety alignment, primarily focusing on preventing harmful content in the initial tokens of the generated output. While methods like resetting can help recover from unsafe generations by discarding previous tokens and restarting the generation","authors_text":"Bilgehan Sel, Dingcheng Li, Ming Jin, Phillip Wallis, Siddhartha Reddy Jonnalagadda, Vaishakh Keshava","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-03-11T22:04:22Z","title":"Backtracking for Safety"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.08919","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":"sha256:3a8d64c985c3b88386bb04c981747f76d524de418329fcc9b380b02fd03b5cf4","target":"record","created_at":"2026-07-05T10:29:43Z","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":"aae3f99d7e455e304af1c70643eae5c4fcd17c5c1c31726bc9e3f3c55d079e62","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-03-11T22:04:22Z","title_canon_sha256":"1d019fba93699be5a9700ba3dcc870ac55a4e6f984e7517be84a3ac6da39a660"},"schema_version":"1.0","source":{"id":"2503.08919","kind":"arxiv","version":1}},"canonical_sha256":"dbddcdec282d957a1356ab9a70c7b8565c14affba9762f1ec12f57ab52b9157b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"dbddcdec282d957a1356ab9a70c7b8565c14affba9762f1ec12f57ab52b9157b","first_computed_at":"2026-07-05T10:29:43.739686Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:29:43.739686Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"JHOyxFbJXd4WO8/BNYuXGnWjURIFpAOpYTVVNcllNSGi9ZUWh9lwafA/QQRLn02+ue4goYiILJ20665NFDAHBA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:29:43.740182Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.08919","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3a8d64c985c3b88386bb04c981747f76d524de418329fcc9b380b02fd03b5cf4","sha256:c35a8639d1958600b4f269c92f15fe977855df8de1dfc2fc35b6b763497bd084"],"state_sha256":"131ac04d3a5bec93c0a66147a657404ecdb0533dbe76eb62b06d140af40e510b"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HW8pJTF7qFFUK3CrN+3kTtqLHBKJUCnxaxTmAbYs9gAKqco7M57fx4MtYi/YwtzDQm8NGcd/6GlTDZUjHAVhDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T11:25:24.488923Z","bundle_sha256":"7220df04e069c0200ef4db1b5065b4b8e42cf48a7808251ec67cd7dce5d2d75f"}}