{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ZRYZW3UCYCCRCR74GWIUGKWWWB","short_pith_number":"pith:ZRYZW3UC","schema_version":"1.0","canonical_sha256":"cc719b6e82c0851147fc3591432ad6b07a8d9d36a4a246488ba220fb17625aee","source":{"kind":"arxiv","id":"2608.04268","version":1},"attestation_state":"computed","paper":{"title":"The Fairness Collapse Phenomenon: Bias Amplification in Language Models Trained on Synthetic Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Antoine Gourru, Irina Proskurina, Julien Velcin","submitted_at":"2026-08-04T22:56:39Z","abstract_excerpt":"Generative models trained on artificially generated data have been shown to exhibit model collapse, resulting in significant performance degradation. As synthetic content increasingly contaminates the training corpora of language models, this raises critical concerns about the use of open data in continued pretraining. Although previous work has demonstrated model collapse in language models, it remains unclear whether exposure to synthetic data amplifies or attenuates the social biases already present in pretrained models. Because language models are known to reproduce and amplify demographic"},"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":"2608.04268","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-08-04T22:56:39Z","cross_cats_sorted":[],"title_canon_sha256":"194e8226702349a8582d26569305017a85f27c3dce5128bd9e8168ce22b5f19f","abstract_canon_sha256":"d8138f6e54630aaaa69a5e559e4bb933adad78de8820c25138b3ed7b30731f29"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-06T01:31:51.835988Z","signature_b64":"GblMeTVFUBq3Mtxdo0NeEjj1p4ur+8DHH2djWgIDywKOdeCyuYEiJyMA1LgoOcASSTO6oWkviuUMR4gtUrytCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cc719b6e82c0851147fc3591432ad6b07a8d9d36a4a246488ba220fb17625aee","last_reissued_at":"2026-08-06T01:31:51.834520Z","signature_status":"signed_v1","first_computed_at":"2026-08-06T01:31:51.834520Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Fairness Collapse Phenomenon: Bias Amplification in Language Models Trained on Synthetic Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Antoine Gourru, Irina Proskurina, Julien Velcin","submitted_at":"2026-08-04T22:56:39Z","abstract_excerpt":"Generative models trained on artificially generated data have been shown to exhibit model collapse, resulting in significant performance degradation. As synthetic content increasingly contaminates the training corpora of language models, this raises critical concerns about the use of open data in continued pretraining. Although previous work has demonstrated model collapse in language models, it remains unclear whether exposure to synthetic data amplifies or attenuates the social biases already present in pretrained models. Because language models are known to reproduce and amplify demographic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.04268","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/2608.04268/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":"2608.04268","created_at":"2026-08-06T01:31:51.836321+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.04268v1","created_at":"2026-08-06T01:31:51.836321+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.04268","created_at":"2026-08-06T01:31:51.836321+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZRYZW3UCYCCR","created_at":"2026-08-06T01:31:51.836321+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZRYZW3UCYCCRCR74","created_at":"2026-08-06T01:31:51.836321+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZRYZW3UC","created_at":"2026-08-06T01:31:51.836321+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/ZRYZW3UCYCCRCR74GWIUGKWWWB","json":"https://pith.science/pith/ZRYZW3UCYCCRCR74GWIUGKWWWB.json","graph_json":"https://pith.science/api/pith-number/ZRYZW3UCYCCRCR74GWIUGKWWWB/graph.json","events_json":"https://pith.science/api/pith-number/ZRYZW3UCYCCRCR74GWIUGKWWWB/events.json","paper":"https://pith.science/paper/ZRYZW3UC"},"agent_actions":{"view_html":"https://pith.science/pith/ZRYZW3UCYCCRCR74GWIUGKWWWB","download_json":"https://pith.science/pith/ZRYZW3UCYCCRCR74GWIUGKWWWB.json","view_paper":"https://pith.science/paper/ZRYZW3UC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.04268&json=true","fetch_graph":"https://pith.science/api/pith-number/ZRYZW3UCYCCRCR74GWIUGKWWWB/graph.json","fetch_events":"https://pith.science/api/pith-number/ZRYZW3UCYCCRCR74GWIUGKWWWB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZRYZW3UCYCCRCR74GWIUGKWWWB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZRYZW3UCYCCRCR74GWIUGKWWWB/action/storage_attestation","attest_author":"https://pith.science/pith/ZRYZW3UCYCCRCR74GWIUGKWWWB/action/author_attestation","sign_citation":"https://pith.science/pith/ZRYZW3UCYCCRCR74GWIUGKWWWB/action/citation_signature","submit_replication":"https://pith.science/pith/ZRYZW3UCYCCRCR74GWIUGKWWWB/action/replication_record"}},"created_at":"2026-08-06T01:31:51.836321+00:00","updated_at":"2026-08-06T01:31:51.836321+00:00"}