{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:APM33HO4CAGKJRV3FOW24EMN63","short_pith_number":"pith:APM33HO4","canonical_record":{"source":{"id":"2508.02063","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-04T05:03:35Z","cross_cats_sorted":[],"title_canon_sha256":"a661c4218ce40746a58eb2e553b12a700e1f29bc647a49d2f6947e460ab34740","abstract_canon_sha256":"9c47888f4127dd1b26b7fb4ea0786060f869c66a9b8954d0d930499da184a2a8"},"schema_version":"1.0"},"canonical_sha256":"03d9bd9ddc100ca4c6bb2badae118df6c299bae56c744b3744631a153f23101f","source":{"kind":"arxiv","id":"2508.02063","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.02063","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"arxiv_version","alias_value":"2508.02063v1","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.02063","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"pith_short_12","alias_value":"APM33HO4CAGK","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"pith_short_16","alias_value":"APM33HO4CAGKJRV3","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"pith_short_8","alias_value":"APM33HO4","created_at":"2026-07-05T11:47:50Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:APM33HO4CAGKJRV3FOW24EMN63","target":"record","payload":{"canonical_record":{"source":{"id":"2508.02063","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-04T05:03:35Z","cross_cats_sorted":[],"title_canon_sha256":"a661c4218ce40746a58eb2e553b12a700e1f29bc647a49d2f6947e460ab34740","abstract_canon_sha256":"9c47888f4127dd1b26b7fb4ea0786060f869c66a9b8954d0d930499da184a2a8"},"schema_version":"1.0"},"canonical_sha256":"03d9bd9ddc100ca4c6bb2badae118df6c299bae56c744b3744631a153f23101f","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:47:50.302660Z","signature_b64":"F4rXoP6lVRSmQxxr18isIKIZfLLmBeNVA363cPJi/4k11vcHXkM/fEpq0cyX1NZCWDwo/qYIKzwyWwsoeHgVBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"03d9bd9ddc100ca4c6bb2badae118df6c299bae56c744b3744631a153f23101f","last_reissued_at":"2026-07-05T11:47:50.302296Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:47:50.302296Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2508.02063","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":"2026-07-05T11:47:50Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"KzF6E3sW4hPhmhBGB+YSMP26m39se3PkTrtijF/I8sWorovPbxN/T0uIeCmwLEwq57a/OZjjIhR8iEsmaM1sAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T19:52:00.729651Z"},"content_sha256":"73b3a6856cf8ee4612bb2169e174a1f40b6c0d7915d0009c460124cb835ae407","schema_version":"1.0","event_id":"sha256:73b3a6856cf8ee4612bb2169e174a1f40b6c0d7915d0009c460124cb835ae407"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:APM33HO4CAGKJRV3FOW24EMN63","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"TRACEALIGN -- Tracing the Drift: Attributing Alignment Failures to Training-Time Belief Sources in LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Aman Chadha, Amitava Das, Vinija Jain","submitted_at":"2025-08-04T05:03:35Z","abstract_excerpt":"Large Language Models (LLMs) fine-tuned to align with human values often exhibit alignment drift, producing unsafe or policy-violating completions when exposed to adversarial prompts, decoding perturbations, or paraphrased jailbreaks. While prior work has behaviorally characterized alignment failure, little is known about the training-time belief sources underlying these failures. We introduce TraceAlign, a unified framework for tracing unsafe completions back to their root causes in the model's training corpus. Central to our approach is the Belief Conflict Index (BCI), which quantifies seman"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.02063","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/2508.02063/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":"2026-07-05T11:47:50Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"L4ThY5fDHj8yll4ae2mFtfGEECQT3ukjIvZlOExui7KwYGUSQtkVvhgXhSJ8Icz0uLo1r+DCZ2aIA2ETt2PvBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T19:52:00.730547Z"},"content_sha256":"ad658f594508db5c7c9e778397f04c288d999a2ee3a80c6db6d57ec9552aace9","schema_version":"1.0","event_id":"sha256:ad658f594508db5c7c9e778397f04c288d999a2ee3a80c6db6d57ec9552aace9"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/APM33HO4CAGKJRV3FOW24EMN63/bundle.json","state_url":"https://pith.science/pith/APM33HO4CAGKJRV3FOW24EMN63/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/APM33HO4CAGKJRV3FOW24EMN63/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":"2026-08-07T19:52:00Z","links":{"resolver":"https://pith.science/pith/APM33HO4CAGKJRV3FOW24EMN63","bundle":"https://pith.science/pith/APM33HO4CAGKJRV3FOW24EMN63/bundle.json","state":"https://pith.science/pith/APM33HO4CAGKJRV3FOW24EMN63/state.json","well_known_bundle":"https://pith.science/.well-known/pith/APM33HO4CAGKJRV3FOW24EMN63/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:APM33HO4CAGKJRV3FOW24EMN63","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":"9c47888f4127dd1b26b7fb4ea0786060f869c66a9b8954d0d930499da184a2a8","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-04T05:03:35Z","title_canon_sha256":"a661c4218ce40746a58eb2e553b12a700e1f29bc647a49d2f6947e460ab34740"},"schema_version":"1.0","source":{"id":"2508.02063","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.02063","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"arxiv_version","alias_value":"2508.02063v1","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.02063","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"pith_short_12","alias_value":"APM33HO4CAGK","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"pith_short_16","alias_value":"APM33HO4CAGKJRV3","created_at":"2026-07-05T11:47:50Z"},{"alias_kind":"pith_short_8","alias_value":"APM33HO4","created_at":"2026-07-05T11:47:50Z"}],"graph_snapshots":[{"event_id":"sha256:ad658f594508db5c7c9e778397f04c288d999a2ee3a80c6db6d57ec9552aace9","target":"graph","created_at":"2026-07-05T11:47:50Z","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/2508.02063/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) fine-tuned to align with human values often exhibit alignment drift, producing unsafe or policy-violating completions when exposed to adversarial prompts, decoding perturbations, or paraphrased jailbreaks. While prior work has behaviorally characterized alignment failure, little is known about the training-time belief sources underlying these failures. We introduce TraceAlign, a unified framework for tracing unsafe completions back to their root causes in the model's training corpus. Central to our approach is the Belief Conflict Index (BCI), which quantifies seman","authors_text":"Aman Chadha, Amitava Das, Vinija Jain","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-04T05:03:35Z","title":"TRACEALIGN -- Tracing the Drift: Attributing Alignment Failures to Training-Time Belief Sources in LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.02063","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":"sha256:73b3a6856cf8ee4612bb2169e174a1f40b6c0d7915d0009c460124cb835ae407","target":"record","created_at":"2026-07-05T11:47:50Z","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":"9c47888f4127dd1b26b7fb4ea0786060f869c66a9b8954d0d930499da184a2a8","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-04T05:03:35Z","title_canon_sha256":"a661c4218ce40746a58eb2e553b12a700e1f29bc647a49d2f6947e460ab34740"},"schema_version":"1.0","source":{"id":"2508.02063","kind":"arxiv","version":1}},"canonical_sha256":"03d9bd9ddc100ca4c6bb2badae118df6c299bae56c744b3744631a153f23101f","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"03d9bd9ddc100ca4c6bb2badae118df6c299bae56c744b3744631a153f23101f","first_computed_at":"2026-07-05T11:47:50.302296Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:47:50.302296Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"F4rXoP6lVRSmQxxr18isIKIZfLLmBeNVA363cPJi/4k11vcHXkM/fEpq0cyX1NZCWDwo/qYIKzwyWwsoeHgVBg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:47:50.302660Z","signed_message":"canonical_sha256_bytes"},"source_id":"2508.02063","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:73b3a6856cf8ee4612bb2169e174a1f40b6c0d7915d0009c460124cb835ae407","sha256:ad658f594508db5c7c9e778397f04c288d999a2ee3a80c6db6d57ec9552aace9"],"state_sha256":"423aee9d7b402dce14855e125ce3804b520670fc36c8169e5bc365b726a77cd9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0B/wsMIvM611mGiF+ZQu19sU4CqV6QEu1MFqH94XYvd9N2feDouzlPMWmaM+PFTbBmmz1F2ejNYJxQeTN3gKAg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T19:52:00.736261Z","bundle_sha256":"cec443013227df824b5506548592bcb34aa3f83b885f19823e6aa46220847b12"}}