{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:GFIFYHZZQLCX67PTXEO7TYD35G","short_pith_number":"pith:GFIFYHZZ","canonical_record":{"source":{"id":"2602.01425","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-02-01T20:18:11Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"6d3b87413fa46fdfaa842ffcf73d08a38d79ff4247ab10e464beac1757024557","abstract_canon_sha256":"031e02220c1b3b424ad7affcbe95747d2aa9e19bf8c6ce964009ae506a68ed95"},"schema_version":"1.0"},"canonical_sha256":"31505c1f3982c57f7df3b91df9e07be9b693fb3f94480e7e556651e8b2d6b847","source":{"kind":"arxiv","id":"2602.01425","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.01425","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"arxiv_version","alias_value":"2602.01425v2","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.01425","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"pith_short_12","alias_value":"GFIFYHZZQLCX","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"pith_short_16","alias_value":"GFIFYHZZQLCX67PT","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"pith_short_8","alias_value":"GFIFYHZZ","created_at":"2026-06-19T16:12:18Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:GFIFYHZZQLCX67PTXEO7TYD35G","target":"record","payload":{"canonical_record":{"source":{"id":"2602.01425","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-02-01T20:18:11Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"6d3b87413fa46fdfaa842ffcf73d08a38d79ff4247ab10e464beac1757024557","abstract_canon_sha256":"031e02220c1b3b424ad7affcbe95747d2aa9e19bf8c6ce964009ae506a68ed95"},"schema_version":"1.0"},"canonical_sha256":"31505c1f3982c57f7df3b91df9e07be9b693fb3f94480e7e556651e8b2d6b847","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:12:18.258882Z","signature_b64":"rznwDMPHzZF7/EawXndWTXhn/5iirdL5WqR5e0jHL1DNf8vcmDNzU/sTIIWpirxGckm3yvO3MvfNP4l9LWPOAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"31505c1f3982c57f7df3b91df9e07be9b693fb3f94480e7e556651e8b2d6b847","last_reissued_at":"2026-06-19T16:12:18.258466Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:12:18.258466Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2602.01425","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-19T16:12:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6sUBC4Ikm2bKGx+on3QyiJac2ar9tRbRUVvbWBgJVaDXPOcnIGfp8vk+Ws5gAGPcZB+ZShnUMDE5+wiwHijIBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T07:13:33.634978Z"},"content_sha256":"bfdb48e10ee6a716a47dadea035edcaa167aa905504d55fab39a237616e4bb1a","schema_version":"1.0","event_id":"sha256:bfdb48e10ee6a716a47dadea035edcaa167aa905504d55fab39a237616e4bb1a"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:GFIFYHZZQLCX67PTXEO7TYD35G","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"One Probe Won't Catch Them All: Towards Targeted Deception Detection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Devina Jain, Joseph Bloom, Satvik Golechha, Shivam Arora, Vikram Natarajan","submitted_at":"2026-02-01T20:18:11Z","abstract_excerpt":"Linear probes are a promising approach for monitoring AI systems for deceptive behaviour. Previous work has shown that a linear classifier trained on a contrastive instruction pair and a simple dataset can achieve good performance. However, these probes exhibit notable failures even in straightforward scenarios, including spurious correlations and false positives on non-deceptive responses. In this paper, we demonstrate that deception detection is inherently heterogeneous: while a single universal probe achieves modest improvements (+0.032 AUC), post-hoc oracle analysis reveals substantially h"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.01425","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/2602.01425/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":"2026-06-19T16:12:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"f5zE+R7XulmwHcnlFr3RVfmTfzbPWeAeEMB+Bfz6+/DWPw3XzUY+qXaTMBjdnM53E9cYt1V6uDerrFidbOjKBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T07:13:33.635513Z"},"content_sha256":"72a39587f6c296d0b801c12cd87379060d71a44ab59f8e585cb8cdb5fe337208","schema_version":"1.0","event_id":"sha256:72a39587f6c296d0b801c12cd87379060d71a44ab59f8e585cb8cdb5fe337208"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/GFIFYHZZQLCX67PTXEO7TYD35G/bundle.json","state_url":"https://pith.science/pith/GFIFYHZZQLCX67PTXEO7TYD35G/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/GFIFYHZZQLCX67PTXEO7TYD35G/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":"2026-08-05T07:13:33Z","links":{"resolver":"https://pith.science/pith/GFIFYHZZQLCX67PTXEO7TYD35G","bundle":"https://pith.science/pith/GFIFYHZZQLCX67PTXEO7TYD35G/bundle.json","state":"https://pith.science/pith/GFIFYHZZQLCX67PTXEO7TYD35G/state.json","well_known_bundle":"https://pith.science/.well-known/pith/GFIFYHZZQLCX67PTXEO7TYD35G/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:GFIFYHZZQLCX67PTXEO7TYD35G","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":"031e02220c1b3b424ad7affcbe95747d2aa9e19bf8c6ce964009ae506a68ed95","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-02-01T20:18:11Z","title_canon_sha256":"6d3b87413fa46fdfaa842ffcf73d08a38d79ff4247ab10e464beac1757024557"},"schema_version":"1.0","source":{"id":"2602.01425","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.01425","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"arxiv_version","alias_value":"2602.01425v2","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.01425","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"pith_short_12","alias_value":"GFIFYHZZQLCX","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"pith_short_16","alias_value":"GFIFYHZZQLCX67PT","created_at":"2026-06-19T16:12:18Z"},{"alias_kind":"pith_short_8","alias_value":"GFIFYHZZ","created_at":"2026-06-19T16:12:18Z"}],"graph_snapshots":[{"event_id":"sha256:72a39587f6c296d0b801c12cd87379060d71a44ab59f8e585cb8cdb5fe337208","target":"graph","created_at":"2026-06-19T16:12:18Z","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/2602.01425/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Linear probes are a promising approach for monitoring AI systems for deceptive behaviour. Previous work has shown that a linear classifier trained on a contrastive instruction pair and a simple dataset can achieve good performance. However, these probes exhibit notable failures even in straightforward scenarios, including spurious correlations and false positives on non-deceptive responses. In this paper, we demonstrate that deception detection is inherently heterogeneous: while a single universal probe achieves modest improvements (+0.032 AUC), post-hoc oracle analysis reveals substantially h","authors_text":"Devina Jain, Joseph Bloom, Satvik Golechha, Shivam Arora, Vikram Natarajan","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-02-01T20:18:11Z","title":"One Probe Won't Catch Them All: Towards Targeted Deception Detection"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.01425","kind":"arxiv","version":2},"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:bfdb48e10ee6a716a47dadea035edcaa167aa905504d55fab39a237616e4bb1a","target":"record","created_at":"2026-06-19T16:12:18Z","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":"031e02220c1b3b424ad7affcbe95747d2aa9e19bf8c6ce964009ae506a68ed95","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-02-01T20:18:11Z","title_canon_sha256":"6d3b87413fa46fdfaa842ffcf73d08a38d79ff4247ab10e464beac1757024557"},"schema_version":"1.0","source":{"id":"2602.01425","kind":"arxiv","version":2}},"canonical_sha256":"31505c1f3982c57f7df3b91df9e07be9b693fb3f94480e7e556651e8b2d6b847","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"31505c1f3982c57f7df3b91df9e07be9b693fb3f94480e7e556651e8b2d6b847","first_computed_at":"2026-06-19T16:12:18.258466Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-19T16:12:18.258466Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"rznwDMPHzZF7/EawXndWTXhn/5iirdL5WqR5e0jHL1DNf8vcmDNzU/sTIIWpirxGckm3yvO3MvfNP4l9LWPOAQ==","signature_status":"signed_v1","signed_at":"2026-06-19T16:12:18.258882Z","signed_message":"canonical_sha256_bytes"},"source_id":"2602.01425","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:bfdb48e10ee6a716a47dadea035edcaa167aa905504d55fab39a237616e4bb1a","sha256:72a39587f6c296d0b801c12cd87379060d71a44ab59f8e585cb8cdb5fe337208"],"state_sha256":"a694586ca01c5236680f66df8eb624ca2c4d8c079bf655c5dc45136696fc199b"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"15ljzkhx077TYsCU5Hr0cVUJUfpe3VGWRo3GJlSiGTOM5FNIPeZECdrMoL1MPyeeDPB2iKarJkOxOW3P5Y6vCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T07:13:33.639528Z","bundle_sha256":"a0662dfa1e99ef92bf3933c6b99e7640968d1d399526db9e254382dc62859e96"}}