{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:IJBMPAA4CWIQGWPJ4MANEG4XJJ","short_pith_number":"pith:IJBMPAA4","canonical_record":{"source":{"id":"2509.04796","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-09-05T04:29:15Z","cross_cats_sorted":[],"title_canon_sha256":"7ec8796e2a9407fafdc1f9984bb8d9f5c62612c3b76521d77442c6f2788eef62","abstract_canon_sha256":"27c1f3978ccbab81c4f1227605a8af2a6ebb38496a280e64d75674f4ff1aef7b"},"schema_version":"1.0"},"canonical_sha256":"4242c7801c15910359e9e300d21b974a529cd048dbb005d5eab962dc9bf7afef","source":{"kind":"arxiv","id":"2509.04796","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.04796","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"arxiv_version","alias_value":"2509.04796v1","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.04796","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"pith_short_12","alias_value":"IJBMPAA4CWIQ","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"pith_short_16","alias_value":"IJBMPAA4CWIQGWPJ","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"pith_short_8","alias_value":"IJBMPAA4","created_at":"2026-07-05T12:05:29Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:IJBMPAA4CWIQGWPJ4MANEG4XJJ","target":"record","payload":{"canonical_record":{"source":{"id":"2509.04796","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-09-05T04:29:15Z","cross_cats_sorted":[],"title_canon_sha256":"7ec8796e2a9407fafdc1f9984bb8d9f5c62612c3b76521d77442c6f2788eef62","abstract_canon_sha256":"27c1f3978ccbab81c4f1227605a8af2a6ebb38496a280e64d75674f4ff1aef7b"},"schema_version":"1.0"},"canonical_sha256":"4242c7801c15910359e9e300d21b974a529cd048dbb005d5eab962dc9bf7afef","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:05:29.859942Z","signature_b64":"Yo9GT3QiaMx+7TX4q1ezl9CByULi3Qeyix52HiRFM77MKanrwf7/6Ay7K104GnAhJ/J/rto2xo3fuyVx71WoBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4242c7801c15910359e9e300d21b974a529cd048dbb005d5eab962dc9bf7afef","last_reissued_at":"2026-07-05T12:05:29.859346Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:05:29.859346Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2509.04796","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":"2026-07-05T12:05:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tewHuZmRVstB/OTLXVJfTEyk9jWJUBMACPJOx9G1uXVWKa3R0t1eF5T7gIBluug/nzH+DO80ur0P+0a8iklwCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T14:00:46.352049Z"},"content_sha256":"b12fc75c1e89588e64b025cf184e865165076f0d080117f0ad0460ac47574db4","schema_version":"1.0","event_id":"sha256:b12fc75c1e89588e64b025cf184e865165076f0d080117f0ad0460ac47574db4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:IJBMPAA4CWIQGWPJ4MANEG4XJJ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Knowledge Collapse in LLMs: When Fluency Survives but Facts Fail under Recursive Synthetic Training","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Adriano Koshiyama, Figarri Keisha, Philip Treleaven, Zekun Wu, Ze Wang","submitted_at":"2025-09-05T04:29:15Z","abstract_excerpt":"Large language models increasingly rely on synthetic data due to human-written content scarcity, yet recursive training on model-generated outputs leads to model collapse, a degenerative process threatening factual reliability. We define knowledge collapse as a distinct three-stage phenomenon where factual accuracy deteriorates while surface fluency persists, creating \"confidently wrong\" outputs that pose critical risks in accuracy-dependent domains. Through controlled experiments with recursive synthetic training, we demonstrate that collapse trajectory and timing depend critically on instruc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.04796","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/2509.04796/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":"2026-07-05T12:05:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"bWejFI4Evtyrc9cIYzUR6kKbTJ6YVSVloSd1NOOxGieZW/hzmq/+3kb+98zs/EKnEpbdtuPyVbdLiWVdBYbiDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T14:00:46.354027Z"},"content_sha256":"f6e68f22ddfe1481b3b236447d674c47811877c8db739aa6a6dfbb973976a4be","schema_version":"1.0","event_id":"sha256:f6e68f22ddfe1481b3b236447d674c47811877c8db739aa6a6dfbb973976a4be"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/IJBMPAA4CWIQGWPJ4MANEG4XJJ/bundle.json","state_url":"https://pith.science/pith/IJBMPAA4CWIQGWPJ4MANEG4XJJ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/IJBMPAA4CWIQGWPJ4MANEG4XJJ/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":"2026-08-17T14:00:46Z","links":{"resolver":"https://pith.science/pith/IJBMPAA4CWIQGWPJ4MANEG4XJJ","bundle":"https://pith.science/pith/IJBMPAA4CWIQGWPJ4MANEG4XJJ/bundle.json","state":"https://pith.science/pith/IJBMPAA4CWIQGWPJ4MANEG4XJJ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/IJBMPAA4CWIQGWPJ4MANEG4XJJ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:IJBMPAA4CWIQGWPJ4MANEG4XJJ","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":"27c1f3978ccbab81c4f1227605a8af2a6ebb38496a280e64d75674f4ff1aef7b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-09-05T04:29:15Z","title_canon_sha256":"7ec8796e2a9407fafdc1f9984bb8d9f5c62612c3b76521d77442c6f2788eef62"},"schema_version":"1.0","source":{"id":"2509.04796","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.04796","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"arxiv_version","alias_value":"2509.04796v1","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.04796","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"pith_short_12","alias_value":"IJBMPAA4CWIQ","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"pith_short_16","alias_value":"IJBMPAA4CWIQGWPJ","created_at":"2026-07-05T12:05:29Z"},{"alias_kind":"pith_short_8","alias_value":"IJBMPAA4","created_at":"2026-07-05T12:05:29Z"}],"graph_snapshots":[{"event_id":"sha256:f6e68f22ddfe1481b3b236447d674c47811877c8db739aa6a6dfbb973976a4be","target":"graph","created_at":"2026-07-05T12:05:29Z","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/2509.04796/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models increasingly rely on synthetic data due to human-written content scarcity, yet recursive training on model-generated outputs leads to model collapse, a degenerative process threatening factual reliability. We define knowledge collapse as a distinct three-stage phenomenon where factual accuracy deteriorates while surface fluency persists, creating \"confidently wrong\" outputs that pose critical risks in accuracy-dependent domains. Through controlled experiments with recursive synthetic training, we demonstrate that collapse trajectory and timing depend critically on instruc","authors_text":"Adriano Koshiyama, Figarri Keisha, Philip Treleaven, Zekun Wu, Ze Wang","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-09-05T04:29:15Z","title":"Knowledge Collapse in LLMs: When Fluency Survives but Facts Fail under Recursive Synthetic Training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.04796","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:b12fc75c1e89588e64b025cf184e865165076f0d080117f0ad0460ac47574db4","target":"record","created_at":"2026-07-05T12:05:29Z","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":"27c1f3978ccbab81c4f1227605a8af2a6ebb38496a280e64d75674f4ff1aef7b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-09-05T04:29:15Z","title_canon_sha256":"7ec8796e2a9407fafdc1f9984bb8d9f5c62612c3b76521d77442c6f2788eef62"},"schema_version":"1.0","source":{"id":"2509.04796","kind":"arxiv","version":1}},"canonical_sha256":"4242c7801c15910359e9e300d21b974a529cd048dbb005d5eab962dc9bf7afef","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4242c7801c15910359e9e300d21b974a529cd048dbb005d5eab962dc9bf7afef","first_computed_at":"2026-07-05T12:05:29.859346Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T12:05:29.859346Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Yo9GT3QiaMx+7TX4q1ezl9CByULi3Qeyix52HiRFM77MKanrwf7/6Ay7K104GnAhJ/J/rto2xo3fuyVx71WoBA==","signature_status":"signed_v1","signed_at":"2026-07-05T12:05:29.859942Z","signed_message":"canonical_sha256_bytes"},"source_id":"2509.04796","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b12fc75c1e89588e64b025cf184e865165076f0d080117f0ad0460ac47574db4","sha256:f6e68f22ddfe1481b3b236447d674c47811877c8db739aa6a6dfbb973976a4be"],"state_sha256":"d57140cd80be0230019af2fe75504f3ba94a9f26ed3e2b97ab4cdb711da571af"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"t1+t5oSO1H76KHohCLgg90Xm3cD7ch9I0JRbtcSX/3ZzZtps5deTDckf/jaOOukwaLrv2QCrD1Neg8T+GzhoDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T14:00:46.364326Z","bundle_sha256":"a39ff6e24537e37c07d8423c8f5e9fdc2408ab6ccef28dbfe72cf069aabc3dee"}}