{"id":"83d3383acaa8d5d705f374de651588248aff835c5fbe8ab05ad4613ad6725481","title":"Adversarial robustness does not entail human-aligned categorization","argument":"The paper by Itay Reznik, Raz Lapid, Moshe Sipper and Tal Golan argues that adversarial robustness alone does not guarantee that image classifiers categorize images as people do. The authors tested four adversarially robust ImageNet classifiers against category choices from 60 participants, using controversial stimuli designed to make robust classifiers disagree. On one set of stimuli, the models predicted individual participants’ choices at near-chance accuracy, while other participants’ majority judgments predicted them much more accurately. On stimuli designed around other model pairs, all four classifiers predicted human responses above chance, but less accurately than other participants’ judgments. The authors report a similar pattern for 17 additional RobustBench models: model-participant agreement was lower than participant-participant agreement on the same images.\n\nThe distinction matters because performance on adversarial robustness does not, in these experiments, amount to matching human categorization. The authors therefore argue that robustness and human alignment should be evaluated as distinct properties of visual recognition systems. Their results concern the stimuli, models and participant judgments tested; the abstract does not establish how the findings extend to other images or settings. An open question is whether this model-human mismatch persists beyond AI","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the arXiv abstract, not the full paper; details of the methods and analyses are unavailable here. The evidence describes controversial stimuli generated to elicit disagreements between classifiers; it does not establish whether the reported mismatch generalizes to other images or settings.","speaker":{"name":"Reznik, Itay, Lapid, Raz, Sipper, Moshe, Golan, Tal","roleAtSource":"Article author"},"source":{"title":"Adversarial robustness does not entail human-aligned categorization","publisher":"arXiv q-bio.NC","format":"Article","url":"https://arxiv.org/abs/2611.00715","publishedAt":"2026-10-31","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors argue that adversarial robustness alone does not guarantee human-aligned categorization.","excerpt":"Our findings show that adversarial robustness alone does not guarantee human-aligned categorization","url":"https://arxiv.org/abs/2611.00715","relationship":"Source-stated"},{"text":"For stimuli optimized to elicit disagreement with another model, the classifiers predicted individual participants’ choices at near-chance accuracy, while other participants’ majority judgments predicted them more accurately.","excerpt":"accuracy in predicting individual participants' single-trial category choices was near chance (48.6-51.3%), whereas majority judgments from other participants predicted responses with 85.0-86.8% accuracy.","url":"https://arxiv.org/abs/2611.00715","relationship":"Source-stated"},{"text":"Across 17 additional RobustBench models, mean model-participant agreement was lower than mean participant-participant agreement on the same images.","excerpt":"Across these 17 models, mean model-participant agreement was 67.3%, versus 81.3% mean participant-participant agreement on the same images.","url":"https://arxiv.org/abs/2611.00715","relationship":"Source-stated"},{"text":"The authors propose evaluating robustness and human alignment as distinct properties of visual recognition systems.","excerpt":"suggest that robustness and human alignment should be evaluated as distinct properties of visual recognition systems.","url":"https://arxiv.org/abs/2611.00715","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.00715","checkedAt":"2026-11-10T06:54:14.420Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Does the model-human mismatch reported here persist beyond the controversial stimuli and model set tested?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Adversarial robustness does not entail human-aligned categorization","summary":"The paper by Itay Reznik, Raz Lapid, Moshe Sipper and Tal Golan argues that adversarial robustness alone does not guarantee that image classifiers categorize images as people do. The authors tested four adversarially robust ImageNet classifiers against category choices from 60 participants, using controversial stimuli designed to make robust classifiers disagree. On one set of stimuli, the models predicted individual participants’ choices at near-chance accuracy, while other participants’ majority judgments predicted them much more accurately. On stimuli designed around other model pairs, all four classifiers predicted human responses above chance, but less accurately than other participants’ judgments. The authors report a similar pattern for 17 additional RobustBench models: model-participant agreement was lower than participant-participant agreement on the same images.\n\nThe distinction matters because performance on adversarial robustness does not, in these experiments, amount to matching human categorization. The authors therefore argue that robustness and human alignment should be evaluated as distinct properties of visual recognition systems. Their results concern the stimuli, models and participant judgments tested; the abstract does not establish how the findings extend to other images or settings. An open question is whether this model-human mismatch persists beyond AI","question":"Does the model-human mismatch reported here persist beyond the controversial stimuli and model set tested?","claims":[{"text":"The authors argue that adversarial robustness alone does not guarantee human-aligned categorization.","excerpt":"Our findings show that adversarial robustness alone does not guarantee human-aligned categorization","support":"source-stated"},{"text":"For stimuli optimized to elicit disagreement with another model, the classifiers predicted individual participants’ choices at near-chance accuracy, while other participants’ majority judgments predicted them more accurately.","excerpt":"accuracy in predicting individual participants' single-trial category choices was near chance (48.6-51.3%), whereas majority judgments from other participants predicted responses with 85.0-86.8% accuracy.","support":"source-stated"},{"text":"Across 17 additional RobustBench models, mean model-participant agreement was lower than mean participant-participant agreement on the same images.","excerpt":"Across these 17 models, mean model-participant agreement was 67.3%, versus 81.3% mean participant-participant agreement on the same images.","support":"source-stated"},{"text":"The authors propose evaluating robustness and human alignment as distinct properties of visual recognition systems.","excerpt":"suggest that robustness and human alignment should be evaluated as distinct properties of visual recognition systems.","support":"source-stated"}],"limitations":["The supplied evidence is the arXiv abstract, not the full paper; details of the methods and analyses are unavailable here.","The evidence describes controversial stimuli generated to elicit disagreements between classifiers; it does not establish whether the reported mismatch generalizes to other images or settings."],"mediaId":"e1f94d57b4b7c23d74b5f8ce86bee2a484b7a07ad091a8f73d5a5c340c057a56","correctionReason":null,"id":"83d3383acaa8d5d705f374de651588248aff835c5fbe8ab05ad4613ad6725481","publicationId":"72b5cf28-a697-485e-875a-99c092c9ebb3","version":1,"sourceUrl":"https://arxiv.org/abs/2611.00715","sourceTitle":"Adversarial robustness does not entail human-aligned categorization","publisher":"arXiv q-bio.NC","publishedAt":null,"originalPublished":{"value":"2026-10-31","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-03T08:51:26.998Z","createdAt":"2026-11-10T06:23:34.829Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"e1f94d57b4b7c23d74b5f8ce86bee2a484b7a07ad091a8f73d5a5c340c057a56","kind":"article","url":"https://arxiv.org/abs/2611.00715","sourceUrl":"https://arxiv.org/abs/2611.00715","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-03T18:45:55.037Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.00715","checkedAt":"2026-11-10T06:54:14.420Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"e1f94d57b4b7c23d74b5f8ce86bee2a484b7a07ad091a8f73d5a5c340c057a56","confidence":0.8300000000000001,"provider":"typesafe","model":"jev-1.13.0","inputHash":"9d365ccfc72525a7fe8a7f0201e1fc150bec5697944f90274cb6583bdd236676","decidedAt":"2026-11-03T18:45:55.651Z"},"provenance":{"textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observationId":"fae0c33584998ba478319e53672160ad2b56ede2ae0825d5040272bc8f661df6","retrievedAt":"2026-11-03T18:45:54.961Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"c5557857-80e4-433c-87a3-ab9eab48fe15"},"reviewer":{"model":"gpt-6-luna","runId":"7a09ad3f-1cd1-44e2-86dc-adb22a174dee"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-10T06:23:34.829Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Reznik, Itay","url":null,"kind":"unknown"},{"name":"Lapid, Raz","url":null,"kind":"unknown"},{"name":"Sipper, Moshe","url":null,"kind":"unknown"},{"name":"Golan, Tal","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Reznik, Itay","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Reznik, Itay","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.00715","locator":"head/meta[name=author|citation_author]"},"assertionId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T06:54:14.420Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","sourceUrl":"https://arxiv.org/abs/2611.00715","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z"}},{"actor":{"name":"Lapid, Raz","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lapid, Raz","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.00715","locator":"head/meta[name=author|citation_author]"},"assertionId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T06:54:14.420Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","sourceUrl":"https://arxiv.org/abs/2611.00715","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z"}},{"actor":{"name":"Sipper, Moshe","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sipper, Moshe","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.00715","locator":"head/meta[name=author|citation_author]"},"assertionId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T06:54:14.420Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","sourceUrl":"https://arxiv.org/abs/2611.00715","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z"}},{"actor":{"name":"Golan, Tal","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Golan, Tal","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.00715","locator":"head/meta[name=author|citation_author]"},"assertionId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T06:54:14.420Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","sourceUrl":"https://arxiv.org/abs/2611.00715","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z"}}],"participationCoverage":{"total":4,"returned":4,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.00715","snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/83d3383acaa8d5d705f374de651588248aff835c5fbe8ab05ad4613ad6725481/source"}},"canonicalUrl":"https://escapemeridian.com/readings/adversarial-robustness-does-not-entail-human-aligned-83d3383a","formats":{"html":"/readings/adversarial-robustness-does-not-entail-human-aligned-83d3383a","markdown":"/readings/adversarial-robustness-does-not-entail-human-aligned-83d3383a.md","json":"/readings/adversarial-robustness-does-not-entail-human-aligned-83d3383a.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"83d3383acaa8d5d705f374de651588248aff835c5fbe8ab05ad4613ad6725481","revision":"72b5cf28-a697-485e-875a-99c092c9ebb3","revisionKind":"publication-id","title":"Adversarial robustness does not entail human-aligned categorization","url":"https://escapemeridian.com/readings/adversarial-robustness-does-not-entail-human-aligned-83d3383a"},"source":{"type":"source","id":"https://arxiv.org/abs/2611.00715","revision":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","revisionKind":"sha256-source-text","title":"Adversarial robustness does not entail human-aligned categorization","url":"https://arxiv.org/abs/2611.00715"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"83d3383acaa8d5d705f374de651588248aff835c5fbe8ab05ad4613ad6725481","title":"Adversarial robustness does not entail human-aligned categorization","argument":"The paper by Itay Reznik, Raz Lapid, Moshe Sipper and Tal Golan argues that adversarial robustness alone does not guarantee that image classifiers categorize images as people do. The authors tested four adversarially robust ImageNet classifiers against category choices from 60 participants, using controversial stimuli designed to make robust classifiers disagree. On one set of stimuli, the models predicted individual participants’ choices at near-chance accuracy, while other participants’ majority judgments predicted them much more accurately. On stimuli designed around other model pairs, all four classifiers predicted human responses above chance, but less accurately than other participants’ judgments. The authors report a similar pattern for 17 additional RobustBench models: model-participant agreement was lower than participant-participant agreement on the same images.\n\nThe distinction matters because performance on adversarial robustness does not, in these experiments, amount to matching human categorization. The authors therefore argue that robustness and human alignment should be evaluated as distinct properties of visual recognition systems. Their results concern the stimuli, models and participant judgments tested; the abstract does not establish how the findings extend to other images or settings. An open question is whether this model-human mismatch persists beyond AI","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the arXiv abstract, not the full paper; details of the methods and analyses are unavailable here. The evidence describes controversial stimuli generated to elicit disagreements between classifiers; it does not establish whether the reported mismatch generalizes to other images or settings.","speaker":{"name":"Reznik, Itay, Lapid, Raz, Sipper, Moshe, Golan, Tal","roleAtSource":"Article author"},"source":{"title":"Adversarial robustness does not entail human-aligned categorization","publisher":"arXiv q-bio.NC","format":"Article","url":"https://arxiv.org/abs/2611.00715","publishedAt":"2026-10-31","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors argue that adversarial robustness alone does not guarantee human-aligned categorization.","excerpt":"Our findings show that adversarial robustness alone does not guarantee human-aligned categorization","url":"https://arxiv.org/abs/2611.00715","relationship":"Source-stated"},{"text":"For stimuli optimized to elicit disagreement with another model, the classifiers predicted individual participants’ choices at near-chance accuracy, while other participants’ majority judgments predicted them more accurately.","excerpt":"accuracy in predicting individual participants' single-trial category choices was near chance (48.6-51.3%), whereas majority judgments from other participants predicted responses with 85.0-86.8% accuracy.","url":"https://arxiv.org/abs/2611.00715","relationship":"Source-stated"},{"text":"Across 17 additional RobustBench models, mean model-participant agreement was lower than mean participant-participant agreement on the same images.","excerpt":"Across these 17 models, mean model-participant agreement was 67.3%, versus 81.3% mean participant-participant agreement on the same images.","url":"https://arxiv.org/abs/2611.00715","relationship":"Source-stated"},{"text":"The authors propose evaluating robustness and human alignment as distinct properties of visual recognition systems.","excerpt":"suggest that robustness and human alignment should be evaluated as distinct properties of visual recognition systems.","url":"https://arxiv.org/abs/2611.00715","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.00715","checkedAt":"2026-11-10T06:54:14.420Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Does the model-human mismatch reported here persist beyond the controversial stimuli and model set tested?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Adversarial robustness does not entail human-aligned categorization","summary":"The paper by Itay Reznik, Raz Lapid, Moshe Sipper and Tal Golan argues that adversarial robustness alone does not guarantee that image classifiers categorize images as people do. The authors tested four adversarially robust ImageNet classifiers against category choices from 60 participants, using controversial stimuli designed to make robust classifiers disagree. On one set of stimuli, the models predicted individual participants’ choices at near-chance accuracy, while other participants’ majority judgments predicted them much more accurately. On stimuli designed around other model pairs, all four classifiers predicted human responses above chance, but less accurately than other participants’ judgments. The authors report a similar pattern for 17 additional RobustBench models: model-participant agreement was lower than participant-participant agreement on the same images.\n\nThe distinction matters because performance on adversarial robustness does not, in these experiments, amount to matching human categorization. The authors therefore argue that robustness and human alignment should be evaluated as distinct properties of visual recognition systems. Their results concern the stimuli, models and participant judgments tested; the abstract does not establish how the findings extend to other images or settings. An open question is whether this model-human mismatch persists beyond AI","question":"Does the model-human mismatch reported here persist beyond the controversial stimuli and model set tested?","claims":[{"text":"The authors argue that adversarial robustness alone does not guarantee human-aligned categorization.","excerpt":"Our findings show that adversarial robustness alone does not guarantee human-aligned categorization","support":"source-stated"},{"text":"For stimuli optimized to elicit disagreement with another model, the classifiers predicted individual participants’ choices at near-chance accuracy, while other participants’ majority judgments predicted them more accurately.","excerpt":"accuracy in predicting individual participants' single-trial category choices was near chance (48.6-51.3%), whereas majority judgments from other participants predicted responses with 85.0-86.8% accuracy.","support":"source-stated"},{"text":"Across 17 additional RobustBench models, mean model-participant agreement was lower than mean participant-participant agreement on the same images.","excerpt":"Across these 17 models, mean model-participant agreement was 67.3%, versus 81.3% mean participant-participant agreement on the same images.","support":"source-stated"},{"text":"The authors propose evaluating robustness and human alignment as distinct properties of visual recognition systems.","excerpt":"suggest that robustness and human alignment should be evaluated as distinct properties of visual recognition systems.","support":"source-stated"}],"limitations":["The supplied evidence is the arXiv abstract, not the full paper; details of the methods and analyses are unavailable here.","The evidence describes controversial stimuli generated to elicit disagreements between classifiers; it does not establish whether the reported mismatch generalizes to other images or settings."],"mediaId":"e1f94d57b4b7c23d74b5f8ce86bee2a484b7a07ad091a8f73d5a5c340c057a56","correctionReason":null,"id":"83d3383acaa8d5d705f374de651588248aff835c5fbe8ab05ad4613ad6725481","publicationId":"72b5cf28-a697-485e-875a-99c092c9ebb3","version":1,"sourceUrl":"https://arxiv.org/abs/2611.00715","sourceTitle":"Adversarial robustness does not entail human-aligned categorization","publisher":"arXiv q-bio.NC","publishedAt":null,"originalPublished":{"value":"2026-10-31","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-03T08:51:26.998Z","createdAt":"2026-11-10T06:23:34.829Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"e1f94d57b4b7c23d74b5f8ce86bee2a484b7a07ad091a8f73d5a5c340c057a56","kind":"article","url":"https://arxiv.org/abs/2611.00715","sourceUrl":"https://arxiv.org/abs/2611.00715","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-03T18:45:55.037Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.00715","checkedAt":"2026-11-10T06:54:14.420Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"e1f94d57b4b7c23d74b5f8ce86bee2a484b7a07ad091a8f73d5a5c340c057a56","confidence":0.8300000000000001,"provider":"typesafe","model":"jev-1.13.0","inputHash":"9d365ccfc72525a7fe8a7f0201e1fc150bec5697944f90274cb6583bdd236676","decidedAt":"2026-11-03T18:45:55.651Z"},"provenance":{"textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observationId":"fae0c33584998ba478319e53672160ad2b56ede2ae0825d5040272bc8f661df6","retrievedAt":"2026-11-03T18:45:54.961Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"c5557857-80e4-433c-87a3-ab9eab48fe15"},"reviewer":{"model":"gpt-6-luna","runId":"7a09ad3f-1cd1-44e2-86dc-adb22a174dee"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-10T06:23:34.829Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Reznik, Itay","url":null,"kind":"unknown"},{"name":"Lapid, Raz","url":null,"kind":"unknown"},{"name":"Sipper, Moshe","url":null,"kind":"unknown"},{"name":"Golan, Tal","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Reznik, Itay","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Reznik, Itay","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.00715","locator":"head/meta[name=author|citation_author]"},"assertionId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T06:54:14.420Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","sourceUrl":"https://arxiv.org/abs/2611.00715","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z"}},{"actor":{"name":"Lapid, Raz","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lapid, Raz","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.00715","locator":"head/meta[name=author|citation_author]"},"assertionId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T06:54:14.420Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","sourceUrl":"https://arxiv.org/abs/2611.00715","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z"}},{"actor":{"name":"Sipper, Moshe","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sipper, Moshe","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.00715","locator":"head/meta[name=author|citation_author]"},"assertionId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T06:54:14.420Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","sourceUrl":"https://arxiv.org/abs/2611.00715","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z"}},{"actor":{"name":"Golan, Tal","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Golan, Tal","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.00715","locator":"head/meta[name=author|citation_author]"},"assertionId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T06:54:14.420Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","sourceUrl":"https://arxiv.org/abs/2611.00715","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z"}}],"participationCoverage":{"total":4,"returned":4,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.00715","snapshotId":"bd312a6cb913207b5ade811aaf58df9ae5e4e91978735269e07125c34b633ea8","contentSha256":"380f6e226d0f77e0fcf1dad3547c1a3e99923c65e783633c11311ebe9d3a0034","textHash":"5cb3a37248c9ddf64c9ac3692d4448c815e2db6d9dfe9a4e6725b7d744544bf6","observedAt":"2026-11-10T06:54:14.420Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/83d3383acaa8d5d705f374de651588248aff835c5fbe8ab05ad4613ad6725481/source"}}}],"audits":[{"publicationId":"72b5cf28-a697-485e-875a-99c092c9ebb3","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading correctly attributes the paper to Itay Reznik, Raz Lapid, Moshe Sipper and Tal Golan. It keeps the central conclusion framed as the authors' argument rather than as established fact or EM opinion.","excerpt":"Authors:Itay Reznik, Raz Lapid, Moshe Sipper, Tal Golan"},{"dimension":"summary","status":"uncertain","explanation":"The final summary sentence is cut off: 'persists beyond AI'. It reads as garbled. Smallest repair: complete it to match the question, e.g. 'beyond the controversial stimuli and models tested'. The central terms 'adversarial robustness', 'controversial stimuli', 'ImageNet', 'RobustBench' and 'near chance' are never explained. A non-specialist cannot tell what robustness means or why it might be expected to produce human-like judgments. Add brief plain-language glosses or term-library tooltips for these. 'On one set of stimuli' is also vague; the source specifies stimuli optimized to elicit disagreement with another model.","excerpt":"We generated 1,008 controversial stimuli by optimizing latent inputs to a pretrained GAN to elicit high-confidence disagreements between pairs of robust ImageNet classifiers."},{"dimension":"date","status":"pass","explanation":"The original publication date, 2026-10-31, matches the arXiv v1 submission date. It is not the later feed date of 2026-11-03.","excerpt":"[Submitted on 31 Oct 2026]"},{"dimension":"support","status":"pass","explanation":"All claims and their excerpts match the abstract. The significance statement is limited to these experiments, and the limitations correctly note that only the abstract is available and that generalization is not established. Minor omission: the abstract also reports weak positive model-human correspondence for the two convolutional models. Mentioning it would preserve nuance, but its absence does not contradict the source.","excerpt":"An analysis incorporating model prediction strength and participants' reported confidence found weak positive model-human correspondence for the two convolutional models."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper. It is the most direct access to the source.","excerpt":null}],"createdAt":"2026-11-10T06:23:34.829Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}