{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:LERQYSXH4ZTHG3EKSFUVCMGFZG","short_pith_number":"pith:LERQYSXH","schema_version":"1.0","canonical_sha256":"59230c4ae7e666736c8a91695130c5c9a67d8e88ad4cd114a89d6b330af27669","source":{"kind":"arxiv","id":"2407.06443","version":2},"attestation_state":"computed","paper":{"title":"Exposing Privacy Gaps: Membership Inference Attack on Preference Data for LLM Alignment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Choon Hui Teo, Hyokun Yun, Qizhang Feng, Santhosh Kumar Kasa, Siva Rajesh Kasa, Sravan Babu Bodapati","submitted_at":"2024-07-08T22:53:23Z","abstract_excerpt":"Large Language Models (LLMs) have seen widespread adoption due to their remarkable natural language capabilities. However, when deploying them in real-world settings, it is important to align LLMs to generate texts according to acceptable human standards. Methods such as Proximal Policy Optimization (PPO) and Direct Preference Optimization (DPO) have enabled significant progress in refining LLMs using human preference data. However, the privacy concerns inherent in utilizing such preference data have yet to be adequately studied. In this paper, we investigate the vulnerability of LLMs aligned "},"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":"2407.06443","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-07-08T22:53:23Z","cross_cats_sorted":[],"title_canon_sha256":"371b415b4d55205592f0900fef257bac739b8acd66a15f1fadaf41abdf3a71ea","abstract_canon_sha256":"87458b87801e33e7cc575f894e65d3dbfdd6700bef8f4ca219774b1c4bd88f20"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:54:34.674197Z","signature_b64":"7rtF0Ie1djNYmKllsEZ06lVI+sG49VhUtrrvDxKzigwwac+Ilh8QUASVTebxeuSboJy0Sa5k7loX1S6fIgplDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"59230c4ae7e666736c8a91695130c5c9a67d8e88ad4cd114a89d6b330af27669","last_reissued_at":"2026-07-05T10:54:34.673753Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:54:34.673753Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Exposing Privacy Gaps: Membership Inference Attack on Preference Data for LLM Alignment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Choon Hui Teo, Hyokun Yun, Qizhang Feng, Santhosh Kumar Kasa, Siva Rajesh Kasa, Sravan Babu Bodapati","submitted_at":"2024-07-08T22:53:23Z","abstract_excerpt":"Large Language Models (LLMs) have seen widespread adoption due to their remarkable natural language capabilities. However, when deploying them in real-world settings, it is important to align LLMs to generate texts according to acceptable human standards. Methods such as Proximal Policy Optimization (PPO) and Direct Preference Optimization (DPO) have enabled significant progress in refining LLMs using human preference data. However, the privacy concerns inherent in utilizing such preference data have yet to be adequately studied. In this paper, we investigate the vulnerability of LLMs aligned "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.06443","kind":"arxiv","version":2},"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/2407.06443/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":"2407.06443","created_at":"2026-07-05T10:54:34.673809+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.06443v2","created_at":"2026-07-05T10:54:34.673809+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.06443","created_at":"2026-07-05T10:54:34.673809+00:00"},{"alias_kind":"pith_short_12","alias_value":"LERQYSXH4ZTH","created_at":"2026-07-05T10:54:34.673809+00:00"},{"alias_kind":"pith_short_16","alias_value":"LERQYSXH4ZTHG3EK","created_at":"2026-07-05T10:54:34.673809+00:00"},{"alias_kind":"pith_short_8","alias_value":"LERQYSXH","created_at":"2026-07-05T10:54:34.673809+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2511.14045","citing_title":"Auditing Data Membership in Reinforcement Learning With Verifiable Rewards","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06423","citing_title":"Pop Quiz Attack: Black-box Membership Inference Attacks Against Large Language Models","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18966","citing_title":"Self-Improving Tabular Language Models via Iterative Reward-Guided Post-Training","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07049","citing_title":"Towards Differentially Private Reinforcement Learning with General Function Approximation","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG","json":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG.json","graph_json":"https://pith.science/api/pith-number/LERQYSXH4ZTHG3EKSFUVCMGFZG/graph.json","events_json":"https://pith.science/api/pith-number/LERQYSXH4ZTHG3EKSFUVCMGFZG/events.json","paper":"https://pith.science/paper/LERQYSXH"},"agent_actions":{"view_html":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG","download_json":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG.json","view_paper":"https://pith.science/paper/LERQYSXH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.06443&json=true","fetch_graph":"https://pith.science/api/pith-number/LERQYSXH4ZTHG3EKSFUVCMGFZG/graph.json","fetch_events":"https://pith.science/api/pith-number/LERQYSXH4ZTHG3EKSFUVCMGFZG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG/action/storage_attestation","attest_author":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG/action/author_attestation","sign_citation":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG/action/citation_signature","submit_replication":"https://pith.science/pith/LERQYSXH4ZTHG3EKSFUVCMGFZG/action/replication_record"}},"created_at":"2026-07-05T10:54:34.673809+00:00","updated_at":"2026-07-05T10:54:34.673809+00:00"}