{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:DEQU2XNTVYHCPJKMSENOA4FJWJ","short_pith_number":"pith:DEQU2XNT","schema_version":"1.0","canonical_sha256":"19214d5db3ae0e27a54c911ae070a9b257c723f2fbb4ded3a449b24c20ab91b3","source":{"kind":"arxiv","id":"2506.23603","version":2},"attestation_state":"computed","paper":{"title":"SoK: Semantic Privacy in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Baihe Ma, Caijun Sun, Chen Li, Guangsheng Yu, Qin Wang, Ren Ping Liu, Wei Ni, Xuelei Qi, Xu Wang, Yanna Jiang, Ying He","submitted_at":"2025-06-30T08:08:15Z","abstract_excerpt":"As Large Language Models (LLMs) are increasingly deployed in sensitive domains, traditional data privacy measures prove inadequate for protecting information that is implicit, contextual, or inferable - what we define as semantic privacy. This Systematization of Knowledge (SoK) introduces a lifecycle-centric framework to analyze how semantic privacy risks emerge across input processing, pretraining, fine-tuning, and alignment stages of LLMs. We categorize key attack vectors and assess how current defenses, such as differential privacy, embedding encryption, edge computing, and unlearning, addr"},"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":"2506.23603","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-06-30T08:08:15Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"aa79e2065cfccc810b5df04a1f4b2ff37736eb35e1cb2595fbc28c208dffd557","abstract_canon_sha256":"8ea6066e926a783fb501d09001cbb34348096295d45a50afabf5cbb5a3a2b476"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:38:08.416998Z","signature_b64":"/7CbrrKs68GF1uKqeyF5Ll6BUBLsZT5b36rRFdEN+b1CGKEHJNg95NNYfRlwYI48aKeM9X7rNhzlmfJZlxLyCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"19214d5db3ae0e27a54c911ae070a9b257c723f2fbb4ded3a449b24c20ab91b3","last_reissued_at":"2026-07-05T11:38:08.416379Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:38:08.416379Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SoK: Semantic Privacy in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Baihe Ma, Caijun Sun, Chen Li, Guangsheng Yu, Qin Wang, Ren Ping Liu, Wei Ni, Xuelei Qi, Xu Wang, Yanna Jiang, Ying He","submitted_at":"2025-06-30T08:08:15Z","abstract_excerpt":"As Large Language Models (LLMs) are increasingly deployed in sensitive domains, traditional data privacy measures prove inadequate for protecting information that is implicit, contextual, or inferable - what we define as semantic privacy. This Systematization of Knowledge (SoK) introduces a lifecycle-centric framework to analyze how semantic privacy risks emerge across input processing, pretraining, fine-tuning, and alignment stages of LLMs. We categorize key attack vectors and assess how current defenses, such as differential privacy, embedding encryption, edge computing, and unlearning, addr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.23603","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/2506.23603/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":"2506.23603","created_at":"2026-07-05T11:38:08.416444+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.23603v2","created_at":"2026-07-05T11:38:08.416444+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.23603","created_at":"2026-07-05T11:38:08.416444+00:00"},{"alias_kind":"pith_short_12","alias_value":"DEQU2XNTVYHC","created_at":"2026-07-05T11:38:08.416444+00:00"},{"alias_kind":"pith_short_16","alias_value":"DEQU2XNTVYHCPJKM","created_at":"2026-07-05T11:38:08.416444+00:00"},{"alias_kind":"pith_short_8","alias_value":"DEQU2XNT","created_at":"2026-07-05T11:38:08.416444+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21059","citing_title":"DEFENGRAPH: Knowledge Graph-Enhanced LLMs for Blue Team Cyber Defense","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2602.20867","citing_title":"SoK: Agentic Skills -- Beyond Tool Use in LLM Agents","ref_index":26,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ","json":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ.json","graph_json":"https://pith.science/api/pith-number/DEQU2XNTVYHCPJKMSENOA4FJWJ/graph.json","events_json":"https://pith.science/api/pith-number/DEQU2XNTVYHCPJKMSENOA4FJWJ/events.json","paper":"https://pith.science/paper/DEQU2XNT"},"agent_actions":{"view_html":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ","download_json":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ.json","view_paper":"https://pith.science/paper/DEQU2XNT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.23603&json=true","fetch_graph":"https://pith.science/api/pith-number/DEQU2XNTVYHCPJKMSENOA4FJWJ/graph.json","fetch_events":"https://pith.science/api/pith-number/DEQU2XNTVYHCPJKMSENOA4FJWJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ/action/storage_attestation","attest_author":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ/action/author_attestation","sign_citation":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ/action/citation_signature","submit_replication":"https://pith.science/pith/DEQU2XNTVYHCPJKMSENOA4FJWJ/action/replication_record"}},"created_at":"2026-07-05T11:38:08.416444+00:00","updated_at":"2026-07-05T11:38:08.416444+00:00"}