{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:GWX5LQSWIQNRJJU6TGMAMCE2DK","short_pith_number":"pith:GWX5LQSW","canonical_record":{"source":{"id":"2508.05469","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-07T15:11:43Z","cross_cats_sorted":["cs.IT","math.IT"],"title_canon_sha256":"f63bbda0467da21521db117d5cf6bd28ba7585450ee8b5740d2afb5112ae82fc","abstract_canon_sha256":"e70c7b7aeceee79bb18742f113955ba86826c00c591beba8aae5dbd05dc83b69"},"schema_version":"1.0"},"canonical_sha256":"35afd5c256441b14a69e999806089a1aa987c1e6078a1c93d530b9b956376e7e","source":{"kind":"arxiv","id":"2508.05469","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.05469","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"arxiv_version","alias_value":"2508.05469v4","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.05469","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"pith_short_12","alias_value":"GWX5LQSWIQNR","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"pith_short_16","alias_value":"GWX5LQSWIQNRJJU6","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"pith_short_8","alias_value":"GWX5LQSW","created_at":"2026-06-23T03:13:46Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:GWX5LQSWIQNRJJU6TGMAMCE2DK","target":"record","payload":{"canonical_record":{"source":{"id":"2508.05469","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-07T15:11:43Z","cross_cats_sorted":["cs.IT","math.IT"],"title_canon_sha256":"f63bbda0467da21521db117d5cf6bd28ba7585450ee8b5740d2afb5112ae82fc","abstract_canon_sha256":"e70c7b7aeceee79bb18742f113955ba86826c00c591beba8aae5dbd05dc83b69"},"schema_version":"1.0"},"canonical_sha256":"35afd5c256441b14a69e999806089a1aa987c1e6078a1c93d530b9b956376e7e","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T03:13:46.603753Z","signature_b64":"R4p9a6bKfbjz/yJzr1eItee7/D4hP6QvIb4DLiE+mg7/9ZJqd2LzcN5xwLWvObtOlNW6DGFhWR2zQkixTsp9CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"35afd5c256441b14a69e999806089a1aa987c1e6078a1c93d530b9b956376e7e","last_reissued_at":"2026-06-23T03:13:46.603272Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T03:13:46.603272Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2508.05469","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-23T03:13:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"fKN+/YG273R/MbQ7GknC4UhQ+eQu7rp9GIB/fkn5TgdGqliziZ7eNTaroeAONDwB7PReziU9KwvcbpRVVSGJDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-11T11:05:29.562707Z"},"content_sha256":"fc2b6451da5dde9febc1e38c7f753a66ec5db7d63eb5e5867bbb051453a84c19","schema_version":"1.0","event_id":"sha256:fc2b6451da5dde9febc1e38c7f753a66ec5db7d63eb5e5867bbb051453a84c19"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:GWX5LQSWIQNRJJU6TGMAMCE2DK","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Let's Measure Information Step-by-Step: AI-Based Evaluation Beyond Vibes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth.","cross_cats":["cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Sanmi Koyejo, Zachary Robertson","submitted_at":"2025-08-07T15:11:43Z","abstract_excerpt":"We evaluate artificial intelligence (AI) systems without ground truth by exploiting a link between strategic gaming and information loss. Building on established information theory, we analyze which mechanisms resist adversarial manipulation. This motivates mutual evaluation, where the overseer is treated as a strategic player estimating mutual information by prompting, making truthful agent reporting an optimal strategy. We show that certain f-divergences, such as total variation distance (TVD), maintain polynomial guarantees under attack, building on an established exponential barrier for es"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Certain f-divergences such as total variation distance maintain polynomial guarantees under attack for mutual information estimation in AI evaluation, with TVD-MI retaining AUC 0.70-0.77 under adversarial attacks while other approaches decay toward chance.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that treating the overseer as a strategic player estimating mutual information by prompting makes truthful agent reporting an optimal strategy, which underpins the claim that the method resists adversarial manipulation without ground truth.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"The paper introduces mutual evaluation with f-divergences like total variation distance to robustly estimate information relationships in AI systems without ground truth, showing better resistance to adversarial attacks than other methods.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"3414d5498b7a321e96edf0fb29c1ca4cf0aa453f744a83c80d3ac4383065e287"},"source":{"id":"2508.05469","kind":"arxiv","version":4},"verdict":{"id":"760b1f5f-804d-4630-85e0-eab79172a198","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-18T23:41:16.558764Z","strongest_claim":"Certain f-divergences such as total variation distance maintain polynomial guarantees under attack for mutual information estimation in AI evaluation, with TVD-MI retaining AUC 0.70-0.77 under adversarial attacks while other approaches decay toward chance.","one_line_summary":"The paper introduces mutual evaluation with f-divergences like total variation distance to robustly estimate information relationships in AI systems without ground truth, showing better resistance to adversarial attacks than other methods.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that treating the overseer as a strategic player estimating mutual information by prompting makes truthful agent reporting an optimal strategy, which underpins the claim that the method resists adversarial manipulation without ground truth.","pith_extraction_headline":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2508.05469/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":2,"snapshot_sha256":"b6d8228ae0fdd687baafdead3d1d16faf3261e616b7d1c1e1d9114fda09251b3"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"760b1f5f-804d-4630-85e0-eab79172a198"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-23T03:13:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"w80YDMBsQe36oBkobcWckZYOZLhqM71anhnfr6tvXe4RaSQUhTtXt3Hbe7OPUWqoux/0+RTSf8jcjXl617VcAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-11T11:05:29.563377Z"},"content_sha256":"353fe53c948c48fdaf16d79ab87da3d4678d3b910502ce09b2ed57b9c9282ae8","schema_version":"1.0","event_id":"sha256:353fe53c948c48fdaf16d79ab87da3d4678d3b910502ce09b2ed57b9c9282ae8"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/bundle.json","state_url":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/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":"2026-08-11T11:05:29Z","links":{"resolver":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK","bundle":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/bundle.json","state":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/state.json","well_known_bundle":"https://pith.science/.well-known/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:GWX5LQSWIQNRJJU6TGMAMCE2DK","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":"e70c7b7aeceee79bb18742f113955ba86826c00c591beba8aae5dbd05dc83b69","cross_cats_sorted":["cs.IT","math.IT"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-07T15:11:43Z","title_canon_sha256":"f63bbda0467da21521db117d5cf6bd28ba7585450ee8b5740d2afb5112ae82fc"},"schema_version":"1.0","source":{"id":"2508.05469","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.05469","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"arxiv_version","alias_value":"2508.05469v4","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.05469","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"pith_short_12","alias_value":"GWX5LQSWIQNR","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"pith_short_16","alias_value":"GWX5LQSWIQNRJJU6","created_at":"2026-06-23T03:13:46Z"},{"alias_kind":"pith_short_8","alias_value":"GWX5LQSW","created_at":"2026-06-23T03:13:46Z"}],"graph_snapshots":[{"event_id":"sha256:353fe53c948c48fdaf16d79ab87da3d4678d3b910502ce09b2ed57b9c9282ae8","target":"graph","created_at":"2026-06-23T03:13:46Z","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":"Certain f-divergences such as total variation distance maintain polynomial guarantees under attack for mutual information estimation in AI evaluation, with TVD-MI retaining AUC 0.70-0.77 under adversarial attacks while other approaches decay toward chance."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The assumption that treating the overseer as a strategic player estimating mutual information by prompting makes truthful agent reporting an optimal strategy, which underpins the claim that the method resists adversarial manipulation without ground truth."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"The paper introduces mutual evaluation with f-divergences like total variation distance to robustly estimate information relationships in AI systems without ground truth, showing better resistance to adversarial attacks than other methods."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth."}],"snapshot_sha256":"3414d5498b7a321e96edf0fb29c1ca4cf0aa453f744a83c80d3ac4383065e287"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b6d8228ae0fdd687baafdead3d1d16faf3261e616b7d1c1e1d9114fda09251b3"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2508.05469/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We evaluate artificial intelligence (AI) systems without ground truth by exploiting a link between strategic gaming and information loss. Building on established information theory, we analyze which mechanisms resist adversarial manipulation. This motivates mutual evaluation, where the overseer is treated as a strategic player estimating mutual information by prompting, making truthful agent reporting an optimal strategy. We show that certain f-divergences, such as total variation distance (TVD), maintain polynomial guarantees under attack, building on an established exponential barrier for es","authors_text":"Sanmi Koyejo, Zachary Robertson","cross_cats":["cs.IT","math.IT"],"headline":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-07T15:11:43Z","title":"Let's Measure Information Step-by-Step: AI-Based Evaluation Beyond Vibes"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.05469","kind":"arxiv","version":4},"verdict":{"created_at":"2026-05-18T23:41:16.558764Z","id":"760b1f5f-804d-4630-85e0-eab79172a198","model_set":{"reader":"grok-4.3"},"one_line_summary":"The paper introduces mutual evaluation with f-divergences like total variation distance to robustly estimate information relationships in AI systems without ground truth, showing better resistance to adversarial attacks than other methods.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth.","strongest_claim":"Certain f-divergences such as total variation distance maintain polynomial guarantees under attack for mutual information estimation in AI evaluation, with TVD-MI retaining AUC 0.70-0.77 under adversarial attacks while other approaches decay toward chance.","weakest_assumption":"The assumption that treating the overseer as a strategic player estimating mutual information by prompting makes truthful agent reporting an optimal strategy, which underpins the claim that the method resists adversarial manipulation without ground truth."}},"verdict_id":"760b1f5f-804d-4630-85e0-eab79172a198"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:fc2b6451da5dde9febc1e38c7f753a66ec5db7d63eb5e5867bbb051453a84c19","target":"record","created_at":"2026-06-23T03:13:46Z","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":"e70c7b7aeceee79bb18742f113955ba86826c00c591beba8aae5dbd05dc83b69","cross_cats_sorted":["cs.IT","math.IT"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-07T15:11:43Z","title_canon_sha256":"f63bbda0467da21521db117d5cf6bd28ba7585450ee8b5740d2afb5112ae82fc"},"schema_version":"1.0","source":{"id":"2508.05469","kind":"arxiv","version":4}},"canonical_sha256":"35afd5c256441b14a69e999806089a1aa987c1e6078a1c93d530b9b956376e7e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"35afd5c256441b14a69e999806089a1aa987c1e6078a1c93d530b9b956376e7e","first_computed_at":"2026-06-23T03:13:46.603272Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-23T03:13:46.603272Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"R4p9a6bKfbjz/yJzr1eItee7/D4hP6QvIb4DLiE+mg7/9ZJqd2LzcN5xwLWvObtOlNW6DGFhWR2zQkixTsp9CQ==","signature_status":"signed_v1","signed_at":"2026-06-23T03:13:46.603753Z","signed_message":"canonical_sha256_bytes"},"source_id":"2508.05469","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:fc2b6451da5dde9febc1e38c7f753a66ec5db7d63eb5e5867bbb051453a84c19","sha256:353fe53c948c48fdaf16d79ab87da3d4678d3b910502ce09b2ed57b9c9282ae8"],"state_sha256":"58dea0576a55b38015df60156e98f9b1df4b076b96ddd99108f62a0306b8e4df"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5Laiv2uLraeiMRa0ddRpvVcSDD4ajh40ZBPjjyroxhWZjAE3UTeme58EAlOoqz0XRcvYSBJS/lG3oWmcNcX3CQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-11T11:05:29.568538Z","bundle_sha256":"ade3c27dc4a4ff0e06683dfec61e3d9220d805aff558b5adb9ef12361f7675cd"}}