{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:PAHVS5GTCCZ3SVFM6RLRF6GUUP","short_pith_number":"pith:PAHVS5GT","canonical_record":{"source":{"id":"2607.05898","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-07T06:51:55Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"b9f6b76acea43209d1c3b0a165a1ad9c700a734bf291c8eff4b6ce5a39e2f2e2","abstract_canon_sha256":"0f89c4ea39ceecd10fd15e395c43335d86e2c376dfcdccf876e80d370ee785c6"},"schema_version":"1.0"},"canonical_sha256":"780f5974d310b3b954acf45712f8d4a3e4a6196f89cfd966544842c459296513","source":{"kind":"arxiv","id":"2607.05898","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.05898","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"arxiv_version","alias_value":"2607.05898v1","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.05898","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"pith_short_12","alias_value":"PAHVS5GTCCZ3","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"pith_short_16","alias_value":"PAHVS5GTCCZ3SVFM","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"pith_short_8","alias_value":"PAHVS5GT","created_at":"2026-07-08T01:18:49Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:PAHVS5GTCCZ3SVFM6RLRF6GUUP","target":"record","payload":{"canonical_record":{"source":{"id":"2607.05898","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-07T06:51:55Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"b9f6b76acea43209d1c3b0a165a1ad9c700a734bf291c8eff4b6ce5a39e2f2e2","abstract_canon_sha256":"0f89c4ea39ceecd10fd15e395c43335d86e2c376dfcdccf876e80d370ee785c6"},"schema_version":"1.0"},"canonical_sha256":"780f5974d310b3b954acf45712f8d4a3e4a6196f89cfd966544842c459296513","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-08T01:18:49.619372Z","signature_b64":"X56BRSl5YgrV8cbsgY/cWWmGXY6fiABgxV4IlSyr8PibrgRQrc2Uh24sPG9/7V1AeMUD+u87LpBZ/4rdkuaFAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"780f5974d310b3b954acf45712f8d4a3e4a6196f89cfd966544842c459296513","last_reissued_at":"2026-07-08T01:18:49.618902Z","signature_status":"signed_v1","first_computed_at":"2026-07-08T01:18:49.618902Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2607.05898","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":"2026-07-08T01:18:49Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5y7uqZc1LhJDyU8MHPCjH1sBbpD2zCuhNXF6AfLk4xMGxAC2JepM9UAMfaGC5b9mYFGZrwsbXx5u0wyRY0gjAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T22:07:54.042956Z"},"content_sha256":"2789e34f07dff202dca8031c16493b5185f8640ffae5cf48fdf96f118ac29494","schema_version":"1.0","event_id":"sha256:2789e34f07dff202dca8031c16493b5185f8640ffae5cf48fdf96f118ac29494"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:PAHVS5GTCCZ3SVFM6RLRF6GUUP","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Auditing of Unlearning Algorithms","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Anastasia Koloskova, Sahasrajit Sarmasarkar, Sanmi Koyejo","submitted_at":"2026-07-07T06:51:55Z","abstract_excerpt":"Evaluating whether unlearning algorithms truly remove training data influence remains an open challenge. We propose a practical auditor that computes data-dependent lower bounds on the unlearning parameter $\\varepsilon$ using membership inference attacks. Evaluating multiple unlearning algorithms, we find a sharp separation: algorithms with rigorous guarantees, such as model clipping and rewind-to-delete, achieve very small $\\varepsilon$ bounds that do not falsify their unlearning guarantees, whereas empirical methods such as Hessian-based unlearning, interleaved ascent-descent, ascent on the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.05898","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/2607.05898/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":"2026-07-08T01:18:49Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"233F515HSdFQgRhdw9DRO/MAGEMuxv5FdHhH33GcjtU4J5Y6AuJoHIcLpbAd9iyk/M7MvEEVO1bs4K4mcRvCAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T22:07:54.043548Z"},"content_sha256":"7cb746f4548e30472ef16a231975f0af7441d5c9348df36e9766877d48e8cf76","schema_version":"1.0","event_id":"sha256:7cb746f4548e30472ef16a231975f0af7441d5c9348df36e9766877d48e8cf76"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/PAHVS5GTCCZ3SVFM6RLRF6GUUP/bundle.json","state_url":"https://pith.science/pith/PAHVS5GTCCZ3SVFM6RLRF6GUUP/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/PAHVS5GTCCZ3SVFM6RLRF6GUUP/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":"2026-08-17T22:07:54Z","links":{"resolver":"https://pith.science/pith/PAHVS5GTCCZ3SVFM6RLRF6GUUP","bundle":"https://pith.science/pith/PAHVS5GTCCZ3SVFM6RLRF6GUUP/bundle.json","state":"https://pith.science/pith/PAHVS5GTCCZ3SVFM6RLRF6GUUP/state.json","well_known_bundle":"https://pith.science/.well-known/pith/PAHVS5GTCCZ3SVFM6RLRF6GUUP/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:PAHVS5GTCCZ3SVFM6RLRF6GUUP","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":"0f89c4ea39ceecd10fd15e395c43335d86e2c376dfcdccf876e80d370ee785c6","cross_cats_sorted":["cs.CR"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-07T06:51:55Z","title_canon_sha256":"b9f6b76acea43209d1c3b0a165a1ad9c700a734bf291c8eff4b6ce5a39e2f2e2"},"schema_version":"1.0","source":{"id":"2607.05898","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.05898","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"arxiv_version","alias_value":"2607.05898v1","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.05898","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"pith_short_12","alias_value":"PAHVS5GTCCZ3","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"pith_short_16","alias_value":"PAHVS5GTCCZ3SVFM","created_at":"2026-07-08T01:18:49Z"},{"alias_kind":"pith_short_8","alias_value":"PAHVS5GT","created_at":"2026-07-08T01:18:49Z"}],"graph_snapshots":[{"event_id":"sha256:7cb746f4548e30472ef16a231975f0af7441d5c9348df36e9766877d48e8cf76","target":"graph","created_at":"2026-07-08T01:18:49Z","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/2607.05898/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Evaluating whether unlearning algorithms truly remove training data influence remains an open challenge. We propose a practical auditor that computes data-dependent lower bounds on the unlearning parameter $\\varepsilon$ using membership inference attacks. Evaluating multiple unlearning algorithms, we find a sharp separation: algorithms with rigorous guarantees, such as model clipping and rewind-to-delete, achieve very small $\\varepsilon$ bounds that do not falsify their unlearning guarantees, whereas empirical methods such as Hessian-based unlearning, interleaved ascent-descent, ascent on the ","authors_text":"Anastasia Koloskova, Sahasrajit Sarmasarkar, Sanmi Koyejo","cross_cats":["cs.CR"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-07T06:51:55Z","title":"Auditing of Unlearning Algorithms"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.05898","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":"sha256:2789e34f07dff202dca8031c16493b5185f8640ffae5cf48fdf96f118ac29494","target":"record","created_at":"2026-07-08T01:18:49Z","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":"0f89c4ea39ceecd10fd15e395c43335d86e2c376dfcdccf876e80d370ee785c6","cross_cats_sorted":["cs.CR"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-07T06:51:55Z","title_canon_sha256":"b9f6b76acea43209d1c3b0a165a1ad9c700a734bf291c8eff4b6ce5a39e2f2e2"},"schema_version":"1.0","source":{"id":"2607.05898","kind":"arxiv","version":1}},"canonical_sha256":"780f5974d310b3b954acf45712f8d4a3e4a6196f89cfd966544842c459296513","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"780f5974d310b3b954acf45712f8d4a3e4a6196f89cfd966544842c459296513","first_computed_at":"2026-07-08T01:18:49.618902Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-08T01:18:49.618902Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"X56BRSl5YgrV8cbsgY/cWWmGXY6fiABgxV4IlSyr8PibrgRQrc2Uh24sPG9/7V1AeMUD+u87LpBZ/4rdkuaFAQ==","signature_status":"signed_v1","signed_at":"2026-07-08T01:18:49.619372Z","signed_message":"canonical_sha256_bytes"},"source_id":"2607.05898","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:2789e34f07dff202dca8031c16493b5185f8640ffae5cf48fdf96f118ac29494","sha256:7cb746f4548e30472ef16a231975f0af7441d5c9348df36e9766877d48e8cf76"],"state_sha256":"4991fe938a6916ba06beedc07e41e9b04f05e49c20d41dfb515e915f8a14cf71"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"AmE/jIDpDeeXDdFyjyFzpYmWjPQ2NNZDONXw9WH0NeAaalwjKGQ32DehsG1tv9iVRi/n0EO78RdZly1BCDoCAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T22:07:54.090464Z","bundle_sha256":"833abb40b32c8fa01b7cda5ec07098f25de65bf5057390142e1c0ab183f2dbd9"}}