{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:N22MO2SYTKU6Y6NWLIHXSBSCZE","short_pith_number":"pith:N22MO2SY","schema_version":"1.0","canonical_sha256":"6eb4c76a589aa9ec79b65a0f790642c915cfff4a19dce747dcab92b03d8bd4bc","source":{"kind":"arxiv","id":"2508.17456","version":2},"attestation_state":"computed","paper":{"title":"Adversarial Examples Are Not Bugs, They Are Superposition","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Liv Gorton, Owen Lewis","submitted_at":"2025-08-24T17:19:53Z","abstract_excerpt":"Adversarial examples -- inputs with imperceptible perturbations that fool neural networks -- remain one of deep learning's most perplexing phenomena despite nearly a decade of research. While numerous defenses and explanations have been proposed, there is no consensus on the fundamental mechanism. One underexplored hypothesis is that superposition, a concept from mechanistic interpretability, may be a major contributing factor, or even the primary cause. We present four lines of evidence in support of this hypothesis, greatly extending prior arguments by Elhage et al. (2022): (1) superposition"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2508.17456","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-24T17:19:53Z","cross_cats_sorted":[],"title_canon_sha256":"6b6f9d0bd37cf073043a065ec3ea03bb9c102a32c8004dcc6d12f9d618c3d22c","abstract_canon_sha256":"309f5fc3763cba44f4f142200e2c1f4ae23a5c2c400c8c7886ff04371ade23ad"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:11:20.095769Z","signature_b64":"HB23KCuB6TWhXks2suJDZK+kFJqOqST9M40mXfjh7o+UtSQ6TrdJUZt+iSzjNNpCDSXIWJKynKcWtzWUAc6WCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6eb4c76a589aa9ec79b65a0f790642c915cfff4a19dce747dcab92b03d8bd4bc","last_reissued_at":"2026-07-05T12:11:20.095276Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:11:20.095276Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Adversarial Examples Are Not Bugs, They Are Superposition","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Liv Gorton, Owen Lewis","submitted_at":"2025-08-24T17:19:53Z","abstract_excerpt":"Adversarial examples -- inputs with imperceptible perturbations that fool neural networks -- remain one of deep learning's most perplexing phenomena despite nearly a decade of research. While numerous defenses and explanations have been proposed, there is no consensus on the fundamental mechanism. One underexplored hypothesis is that superposition, a concept from mechanistic interpretability, may be a major contributing factor, or even the primary cause. We present four lines of evidence in support of this hypothesis, greatly extending prior arguments by Elhage et al. (2022): (1) superposition"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.17456","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/2508.17456/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2508.17456","created_at":"2026-07-05T12:11:20.095345+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.17456v2","created_at":"2026-07-05T12:11:20.095345+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.17456","created_at":"2026-07-05T12:11:20.095345+00:00"},{"alias_kind":"pith_short_12","alias_value":"N22MO2SYTKU6","created_at":"2026-07-05T12:11:20.095345+00:00"},{"alias_kind":"pith_short_16","alias_value":"N22MO2SYTKU6Y6NW","created_at":"2026-07-05T12:11:20.095345+00:00"},{"alias_kind":"pith_short_8","alias_value":"N22MO2SY","created_at":"2026-07-05T12:11:20.095345+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE","json":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE.json","graph_json":"https://pith.science/api/pith-number/N22MO2SYTKU6Y6NWLIHXSBSCZE/graph.json","events_json":"https://pith.science/api/pith-number/N22MO2SYTKU6Y6NWLIHXSBSCZE/events.json","paper":"https://pith.science/paper/N22MO2SY"},"agent_actions":{"view_html":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE","download_json":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE.json","view_paper":"https://pith.science/paper/N22MO2SY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.17456&json=true","fetch_graph":"https://pith.science/api/pith-number/N22MO2SYTKU6Y6NWLIHXSBSCZE/graph.json","fetch_events":"https://pith.science/api/pith-number/N22MO2SYTKU6Y6NWLIHXSBSCZE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE/action/storage_attestation","attest_author":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE/action/author_attestation","sign_citation":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE/action/citation_signature","submit_replication":"https://pith.science/pith/N22MO2SYTKU6Y6NWLIHXSBSCZE/action/replication_record"}},"created_at":"2026-07-05T12:11:20.095345+00:00","updated_at":"2026-07-05T12:11:20.095345+00:00"}