{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:BWA3S2Y3ITO6EYF2RZLBDD6I4U","short_pith_number":"pith:BWA3S2Y3","canonical_record":{"source":{"id":"2605.15455","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.HC","submitted_at":"2026-05-14T22:37:56Z","cross_cats_sorted":[],"title_canon_sha256":"4e94b40a85ef2f6b8f2dff3a0f4fe9b1e5722ded430fba11594ca3e9ef6a6ac2","abstract_canon_sha256":"cef7bff4139af60dbcbf57fe5378fa7d7eba972823e1f5c43915982cc740dddf"},"schema_version":"1.0"},"canonical_sha256":"0d81b96b1b44dde260ba8e56118fc8e519a543a5606796f9d4a8229d90d9af53","source":{"kind":"arxiv","id":"2605.15455","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15455","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15455v1","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15455","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"pith_short_12","alias_value":"BWA3S2Y3ITO6","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"pith_short_16","alias_value":"BWA3S2Y3ITO6EYF2","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"pith_short_8","alias_value":"BWA3S2Y3","created_at":"2026-05-20T00:00:59Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:BWA3S2Y3ITO6EYF2RZLBDD6I4U","target":"record","payload":{"canonical_record":{"source":{"id":"2605.15455","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.HC","submitted_at":"2026-05-14T22:37:56Z","cross_cats_sorted":[],"title_canon_sha256":"4e94b40a85ef2f6b8f2dff3a0f4fe9b1e5722ded430fba11594ca3e9ef6a6ac2","abstract_canon_sha256":"cef7bff4139af60dbcbf57fe5378fa7d7eba972823e1f5c43915982cc740dddf"},"schema_version":"1.0"},"canonical_sha256":"0d81b96b1b44dde260ba8e56118fc8e519a543a5606796f9d4a8229d90d9af53","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:59.478913Z","signature_b64":"zYpfvYvxpjGm8Q+5fgujS+gf5VOF1s13v2b8KzQ7RMBXiPqou/F811Kfib5S95F35bW0BLe6AiDA18sU8D6FDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0d81b96b1b44dde260ba8e56118fc8e519a543a5606796f9d4a8229d90d9af53","last_reissued_at":"2026-05-20T00:00:59.478264Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:59.478264Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.15455","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":"2026-05-20T00:00:59Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"V1woh7JUSTpKnHR1VpUN2pThfvIJBPx/jaO7Lc5hhNnedTe7q0+1uXjeaz3FObP1RvxoDrfCkw2XJTozP3+2Dg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T10:15:40.970724Z"},"content_sha256":"c7e8b0621f2a25dfcfb03ac131d77d0dbe292ba798c6f957c15fb8202ce243ce","schema_version":"1.0","event_id":"sha256:c7e8b0621f2a25dfcfb03ac131d77d0dbe292ba798c6f957c15fb8202ce243ce"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:BWA3S2Y3ITO6EYF2RZLBDD6I4U","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Multi-Turn Neural Transparency: Surfacing Neural Activations Improves User Calibration to LLM Behavioral Drift","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Surfacing an LLM's internal neural activations in real time helps users better anticipate and evaluate shifts in chatbot behavior across a conversation.","cross_cats":[],"primary_cat":"cs.HC","authors_text":"Anthony Baez, Pat Pataranutaporn, Sheer Karny","submitted_at":"2026-05-14T22:37:56Z","abstract_excerpt":"Chatbot behavior is often opaque to users, as responses can shift unpredictably across a conversation, drifting toward sycophancy, toxicity, or other unsafe responses. This can leave users vulnerable, either being misled by overly agreeable AI or manipulated by a harmful chatbot that no longer behaves as intended. To address this, we introduce multi-turn neural transparency, an interface that surfaces an LLM's internal neural activations in real time to help users anticipate and recognize how behaviors change across turns. We construct behavioral vectors for six personality traits using method"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Participants without visualization struggled to accurately evaluate traits (RMSE ≈ 0.6-0.7), while the inclusion of neural transparency significantly improved both anticipation and evaluation compared to no visualization (d = -0.34 to -0.49). The multi-turn dynamic visualization additionally outperformed the static single-turn visualization on holistic evaluation of model behavior (d = -0.32). Transparency also reduced overconfidence.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The behavioral vectors identified via contrastive system prompts accurately and stably represent the expression of the six personality traits in the LLM's activation space across different models and contexts.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Multi-turn neural transparency using behavioral vectors and dynamic visualizations improves user anticipation and evaluation of LLM trait expression while reducing overconfidence, per a randomized study with 246 participants.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Surfacing an LLM's internal neural activations in real time helps users better anticipate and evaluate shifts in chatbot behavior across a conversation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"835d830a4446577f1f26f71a082d2a4e24385deb0180f7f5b5459c81dd02b958"},"source":{"id":"2605.15455","kind":"arxiv","version":1},"verdict":{"id":"aef57067-dbfc-43d0-98ac-e7c633e9ec67","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T14:33:32.517835Z","strongest_claim":"Participants without visualization struggled to accurately evaluate traits (RMSE ≈ 0.6-0.7), while the inclusion of neural transparency significantly improved both anticipation and evaluation compared to no visualization (d = -0.34 to -0.49). The multi-turn dynamic visualization additionally outperformed the static single-turn visualization on holistic evaluation of model behavior (d = -0.32). Transparency also reduced overconfidence.","one_line_summary":"Multi-turn neural transparency using behavioral vectors and dynamic visualizations improves user anticipation and evaluation of LLM trait expression while reducing overconfidence, per a randomized study with 246 participants.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The behavioral vectors identified via contrastive system prompts accurately and stably represent the expression of the six personality traits in the LLM's activation space across different models and contexts.","pith_extraction_headline":"Surfacing an LLM's internal neural activations in real time helps users better anticipate and evaluate shifts in chatbot behavior across a conversation."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15455/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"citation_quote_validity","ran_at":"2026-05-19T15:49:50.640823Z","status":"completed","version":"0.1.0","findings_count":0},{"name":"cited_work_retraction","ran_at":"2026-05-19T15:23:32.955604Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T15:01:17.608677Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T14:50:17.023442Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T14:21:54.105065Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T13:33:22.674582Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"327a8db5a078d8332fcc38a7d1143bdb08fb9a1f55e3f6413d5307ccd76f1421"},"references":{"count":52,"sample":[{"doi":"","year":2016,"title":"Understanding intermediate layers using linear classifier probes","work_id":"bdc944db-4be2-44f7-950b-eaef12fab00e","ref_index":1,"cited_arxiv_id":"1610.01644","is_internal_anchor":true},{"doi":"","year":2025,"title":"Emmanuel Ameisen, Jack Lindsey, Adam Pearce, Wes Gurnee, Nicholas L. Turner, Brian Chen, Craig Citro, David Abrahams, Shan Carter, Basil Hosmer, Jonathan Marcus, Michael Sklar, Adly Templeton, Trenton","work_id":"fcae98cd-5e7a-4d2a-a933-79c40c3c92f0","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Chayapatr Archiwaranguprok, Constanze Albrecht, Pattie Maes, Karrie Kara- halios, and Pat Pataranutaporn. 2025. Simulating Psychological Risks in Human- AI Interactions: Real-Case Informed Modeling of","work_id":"a529f005-83d4-49ed-8d06-79bbfea26eda","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Andy Arditi, Oscar Obeso, Aaquib Syed, Daniel Paleka, Nina Panickssery, Wes Gurnee, and Neel Nanda. 2024. Refusal in language models is mediated by a single direction.Advances in Neural Information Pr","work_id":"a1aed2c4-fc25-430f-9e8c-8517c52cd1f8","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Persona Vectors: Monitoring and Controlling Character Traits in Language Models","work_id":"cf32dbef-9132-4648-abcb-0ebf3ac3af80","ref_index":5,"cited_arxiv_id":"2507.21509","is_internal_anchor":true}],"resolved_work":52,"snapshot_sha256":"837ac3bbfccecf981662fe6c6a64455eae9f0a870c1783dcf5672d1e9745bef9","internal_anchors":6},"formal_canon":{"evidence_count":1,"snapshot_sha256":"30ed98126631d5c9ab0e2a25138599ebf5ae9dbefd7aabb1155254d879c7689e"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"aef57067-dbfc-43d0-98ac-e7c633e9ec67"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:59Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4m5OBmvgBvvFgOKiIxNOEpFB905dTpkyzAcJHZwaOI+fzMHItLJ6DP7gxZGf6UHUoK4JMyCh9cHFiFRo33cdCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T10:15:40.971822Z"},"content_sha256":"b919074a32096c326abb95c55d1c552953dec8f5875447b33950346b1bc530e2","schema_version":"1.0","event_id":"sha256:b919074a32096c326abb95c55d1c552953dec8f5875447b33950346b1bc530e2"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/BWA3S2Y3ITO6EYF2RZLBDD6I4U/bundle.json","state_url":"https://pith.science/pith/BWA3S2Y3ITO6EYF2RZLBDD6I4U/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/BWA3S2Y3ITO6EYF2RZLBDD6I4U/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":"2026-05-27T10:15:40Z","links":{"resolver":"https://pith.science/pith/BWA3S2Y3ITO6EYF2RZLBDD6I4U","bundle":"https://pith.science/pith/BWA3S2Y3ITO6EYF2RZLBDD6I4U/bundle.json","state":"https://pith.science/pith/BWA3S2Y3ITO6EYF2RZLBDD6I4U/state.json","well_known_bundle":"https://pith.science/.well-known/pith/BWA3S2Y3ITO6EYF2RZLBDD6I4U/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:BWA3S2Y3ITO6EYF2RZLBDD6I4U","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":"cef7bff4139af60dbcbf57fe5378fa7d7eba972823e1f5c43915982cc740dddf","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.HC","submitted_at":"2026-05-14T22:37:56Z","title_canon_sha256":"4e94b40a85ef2f6b8f2dff3a0f4fe9b1e5722ded430fba11594ca3e9ef6a6ac2"},"schema_version":"1.0","source":{"id":"2605.15455","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15455","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15455v1","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15455","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"pith_short_12","alias_value":"BWA3S2Y3ITO6","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"pith_short_16","alias_value":"BWA3S2Y3ITO6EYF2","created_at":"2026-05-20T00:00:59Z"},{"alias_kind":"pith_short_8","alias_value":"BWA3S2Y3","created_at":"2026-05-20T00:00:59Z"}],"graph_snapshots":[{"event_id":"sha256:b919074a32096c326abb95c55d1c552953dec8f5875447b33950346b1bc530e2","target":"graph","created_at":"2026-05-20T00:00:59Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Participants without visualization struggled to accurately evaluate traits (RMSE ≈ 0.6-0.7), while the inclusion of neural transparency significantly improved both anticipation and evaluation compared to no visualization (d = -0.34 to -0.49). The multi-turn dynamic visualization additionally outperformed the static single-turn visualization on holistic evaluation of model behavior (d = -0.32). Transparency also reduced overconfidence."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The behavioral vectors identified via contrastive system prompts accurately and stably represent the expression of the six personality traits in the LLM's activation space across different models and contexts."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Multi-turn neural transparency using behavioral vectors and dynamic visualizations improves user anticipation and evaluation of LLM trait expression while reducing overconfidence, per a randomized study with 246 participants."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Surfacing an LLM's internal neural activations in real time helps users better anticipate and evaluate shifts in chatbot behavior across a conversation."}],"snapshot_sha256":"835d830a4446577f1f26f71a082d2a4e24385deb0180f7f5b5459c81dd02b958"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"30ed98126631d5c9ab0e2a25138599ebf5ae9dbefd7aabb1155254d879c7689e"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"citation_quote_validity","ran_at":"2026-05-19T15:49:50.640823Z","status":"completed","version":"0.1.0"},{"findings_count":0,"name":"cited_work_retraction","ran_at":"2026-05-19T15:23:32.955604Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-19T15:01:17.608677Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T14:50:17.023442Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"claim_evidence","ran_at":"2026-05-19T14:21:54.105065Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-19T13:33:22.674582Z","status":"skipped","version":"1.0.0"}],"endpoint":"/pith/2605.15455/integrity.json","findings":[],"snapshot_sha256":"327a8db5a078d8332fcc38a7d1143bdb08fb9a1f55e3f6413d5307ccd76f1421","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Chatbot behavior is often opaque to users, as responses can shift unpredictably across a conversation, drifting toward sycophancy, toxicity, or other unsafe responses. This can leave users vulnerable, either being misled by overly agreeable AI or manipulated by a harmful chatbot that no longer behaves as intended. To address this, we introduce multi-turn neural transparency, an interface that surfaces an LLM's internal neural activations in real time to help users anticipate and recognize how behaviors change across turns. We construct behavioral vectors for six personality traits using method","authors_text":"Anthony Baez, Pat Pataranutaporn, Sheer Karny","cross_cats":[],"headline":"Surfacing an LLM's internal neural activations in real time helps users better anticipate and evaluate shifts in chatbot behavior across a conversation.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.HC","submitted_at":"2026-05-14T22:37:56Z","title":"Multi-Turn Neural Transparency: Surfacing Neural Activations Improves User Calibration to LLM Behavioral Drift"},"references":{"count":52,"internal_anchors":6,"resolved_work":52,"sample":[{"cited_arxiv_id":"1610.01644","doi":"","is_internal_anchor":true,"ref_index":1,"title":"Understanding intermediate layers using linear classifier probes","work_id":"bdc944db-4be2-44f7-950b-eaef12fab00e","year":2016},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Emmanuel Ameisen, Jack Lindsey, Adam Pearce, Wes Gurnee, Nicholas L. Turner, Brian Chen, Craig Citro, David Abrahams, Shan Carter, Basil Hosmer, Jonathan Marcus, Michael Sklar, Adly Templeton, Trenton","work_id":"fcae98cd-5e7a-4d2a-a933-79c40c3c92f0","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Chayapatr Archiwaranguprok, Constanze Albrecht, Pattie Maes, Karrie Kara- halios, and Pat Pataranutaporn. 2025. Simulating Psychological Risks in Human- AI Interactions: Real-Case Informed Modeling of","work_id":"a529f005-83d4-49ed-8d06-79bbfea26eda","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Andy Arditi, Oscar Obeso, Aaquib Syed, Daniel Paleka, Nina Panickssery, Wes Gurnee, and Neel Nanda. 2024. Refusal in language models is mediated by a single direction.Advances in Neural Information Pr","work_id":"a1aed2c4-fc25-430f-9e8c-8517c52cd1f8","year":2024},{"cited_arxiv_id":"2507.21509","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Persona Vectors: Monitoring and Controlling Character Traits in Language Models","work_id":"cf32dbef-9132-4648-abcb-0ebf3ac3af80","year":2025}],"snapshot_sha256":"837ac3bbfccecf981662fe6c6a64455eae9f0a870c1783dcf5672d1e9745bef9"},"source":{"id":"2605.15455","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-19T14:33:32.517835Z","id":"aef57067-dbfc-43d0-98ac-e7c633e9ec67","model_set":{"reader":"grok-4.3"},"one_line_summary":"Multi-turn neural transparency using behavioral vectors and dynamic visualizations improves user anticipation and evaluation of LLM trait expression while reducing overconfidence, per a randomized study with 246 participants.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Surfacing an LLM's internal neural activations in real time helps users better anticipate and evaluate shifts in chatbot behavior across a conversation.","strongest_claim":"Participants without visualization struggled to accurately evaluate traits (RMSE ≈ 0.6-0.7), while the inclusion of neural transparency significantly improved both anticipation and evaluation compared to no visualization (d = -0.34 to -0.49). The multi-turn dynamic visualization additionally outperformed the static single-turn visualization on holistic evaluation of model behavior (d = -0.32). Transparency also reduced overconfidence.","weakest_assumption":"The behavioral vectors identified via contrastive system prompts accurately and stably represent the expression of the six personality traits in the LLM's activation space across different models and contexts."}},"verdict_id":"aef57067-dbfc-43d0-98ac-e7c633e9ec67"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c7e8b0621f2a25dfcfb03ac131d77d0dbe292ba798c6f957c15fb8202ce243ce","target":"record","created_at":"2026-05-20T00:00:59Z","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":"cef7bff4139af60dbcbf57fe5378fa7d7eba972823e1f5c43915982cc740dddf","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.HC","submitted_at":"2026-05-14T22:37:56Z","title_canon_sha256":"4e94b40a85ef2f6b8f2dff3a0f4fe9b1e5722ded430fba11594ca3e9ef6a6ac2"},"schema_version":"1.0","source":{"id":"2605.15455","kind":"arxiv","version":1}},"canonical_sha256":"0d81b96b1b44dde260ba8e56118fc8e519a543a5606796f9d4a8229d90d9af53","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"0d81b96b1b44dde260ba8e56118fc8e519a543a5606796f9d4a8229d90d9af53","first_computed_at":"2026-05-20T00:00:59.478264Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:00:59.478264Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"zYpfvYvxpjGm8Q+5fgujS+gf5VOF1s13v2b8KzQ7RMBXiPqou/F811Kfib5S95F35bW0BLe6AiDA18sU8D6FDA==","signature_status":"signed_v1","signed_at":"2026-05-20T00:00:59.478913Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.15455","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c7e8b0621f2a25dfcfb03ac131d77d0dbe292ba798c6f957c15fb8202ce243ce","sha256:b919074a32096c326abb95c55d1c552953dec8f5875447b33950346b1bc530e2"],"state_sha256":"430d95f82b42339f94f7a0e131ae7c7eea0d4fb7bfe90fe225c38dcbebc2992d"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"AWRtb57R+aK4koI39e5WfmHnvK3fRn0lQhWViNl8xf4Pt2TW0r0FDdvfHRlXp88Esw1uywaHtmFeYE0Gom41Cg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-27T10:15:40.976951Z","bundle_sha256":"882306452440b4d6d23bec5491164fadb188f1eecaea7d98a397729987bce0e2"}}