{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KA7ICOTDHZL4TITMUSLHMEBL3D","short_pith_number":"pith:KA7ICOTD","schema_version":"1.0","canonical_sha256":"503e813a633e57c9a26ca49676102bd8f47736e4df6975fe74e80cc419cbfb1e","source":{"kind":"arxiv","id":"2512.21815","version":3},"attestation_state":"computed","paper":{"title":"High-Entropy Tokens as Multimodal Failure Points in Vision-Language Models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"A small share of high-entropy tokens during generation concentrates most adversarial influence in vision-language models.","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Jing Zhang, Jinhong Ni, Mengqi He, Shu Zou, Xin Shen, Xinyu Tian, Zhaoyuan Yang","submitted_at":"2025-12-26T01:01:25Z","abstract_excerpt":"Vision-language models (VLMs) achieve remarkable performance but remain vulnerable to adversarial attacks. Entropy, as a measure of model uncertainty, is highly correlated with VLM reliability. While prior entropy-based attacks maximize uncertainty at all decoding steps, implicitly assuming that every token equally contributes to model instability, we reveal that a small fraction (around 20%) of high-entropy tokens, in the evaluated representative open-source VLMs with diverse architectures, concentrates a disproportionate share of adversarial influence during autoregressive generation. We dem"},"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":true},"canonical_record":{"source":{"id":"2512.21815","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2025-12-26T01:01:25Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"48f6abf74e09a01bb218402c0ee098c18be6a07cd82f4b547d10d39545aea293","abstract_canon_sha256":"87fcc8f582da8f4dfb1a9eb1534ecb945a80f771a6a8095ecc6320410f5b1e7f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:03:21.257956Z","signature_b64":"sUUHHuqXZzX/fRx7tA/543sZzUuioYa6M8szCh529g8uOH/PV87V8HAyctQv6mfOGrXYtE4klpORRaJslQZ+Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"503e813a633e57c9a26ca49676102bd8f47736e4df6975fe74e80cc419cbfb1e","last_reissued_at":"2026-05-26T01:03:21.256922Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:03:21.256922Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"High-Entropy Tokens as Multimodal Failure Points in Vision-Language Models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"A small share of high-entropy tokens during generation concentrates most adversarial influence in vision-language models.","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Jing Zhang, Jinhong Ni, Mengqi He, Shu Zou, Xin Shen, Xinyu Tian, Zhaoyuan Yang","submitted_at":"2025-12-26T01:01:25Z","abstract_excerpt":"Vision-language models (VLMs) achieve remarkable performance but remain vulnerable to adversarial attacks. Entropy, as a measure of model uncertainty, is highly correlated with VLM reliability. While prior entropy-based attacks maximize uncertainty at all decoding steps, implicitly assuming that every token equally contributes to model instability, we reveal that a small fraction (around 20%) of high-entropy tokens, in the evaluated representative open-source VLMs with diverse architectures, concentrates a disproportionate share of adversarial influence during autoregressive generation. We dem"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"a small fraction (around 20%) of high-entropy tokens, in the evaluated representative open-source VLMs with diverse architectures, concentrates a disproportionate share of adversarial influence during autoregressive generation","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That high-entropy tokens can be identified reliably during generation and that concentrating perturbations on them produces comparable semantic degradation to global attacks without requiring post-hoc selection or model-specific tuning that would invalidate the transferability claim.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"High-entropy tokens act as concentrated multimodal failure points in VLMs, enabling sparse Entropy-Guided Attacks that achieve 93-95% success and 30-38% harmful rates with cross-model transfer.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A small share of high-entropy tokens during generation concentrates most adversarial influence in vision-language models.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"1f8077993881cfb232972a684346efbb0dc5737bc45ee675bedbd820673b7b48"},"source":{"id":"2512.21815","kind":"arxiv","version":3},"verdict":{"id":"ec215b08-5ced-4a3d-a80c-8d4baa9b633b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T19:26:18.378293Z","strongest_claim":"a small fraction (around 20%) of high-entropy tokens, in the evaluated representative open-source VLMs with diverse architectures, concentrates a disproportionate share of adversarial influence during autoregressive generation","one_line_summary":"High-entropy tokens act as concentrated multimodal failure points in VLMs, enabling sparse Entropy-Guided Attacks that achieve 93-95% success and 30-38% harmful rates with cross-model transfer.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That high-entropy tokens can be identified reliably during generation and that concentrating perturbations on them produces comparable semantic degradation to global attacks without requiring post-hoc selection or model-specific tuning that would invalidate the transferability claim.","pith_extraction_headline":"A small share of high-entropy tokens during generation concentrates most adversarial influence in vision-language models."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2512.21815/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":"b318806ae64168b412aa3aabbb7615d7f56ade2d02e8bd896fead889c1bbbaef"},"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":"2512.21815","created_at":"2026-05-26T01:03:21.257084+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.21815v3","created_at":"2026-05-26T01:03:21.257084+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.21815","created_at":"2026-05-26T01:03:21.257084+00:00"},{"alias_kind":"pith_short_12","alias_value":"KA7ICOTDHZL4","created_at":"2026-05-26T01:03:21.257084+00:00"},{"alias_kind":"pith_short_16","alias_value":"KA7ICOTDHZL4TITM","created_at":"2026-05-26T01:03:21.257084+00:00"},{"alias_kind":"pith_short_8","alias_value":"KA7ICOTD","created_at":"2026-05-26T01:03:21.257084+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.10764","citing_title":"Break the Brake, Not the Wheel: Untargeted Jailbreak via Entropy Maximization","ref_index":9,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D","json":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D.json","graph_json":"https://pith.science/api/pith-number/KA7ICOTDHZL4TITMUSLHMEBL3D/graph.json","events_json":"https://pith.science/api/pith-number/KA7ICOTDHZL4TITMUSLHMEBL3D/events.json","paper":"https://pith.science/paper/KA7ICOTD"},"agent_actions":{"view_html":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D","download_json":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D.json","view_paper":"https://pith.science/paper/KA7ICOTD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.21815&json=true","fetch_graph":"https://pith.science/api/pith-number/KA7ICOTDHZL4TITMUSLHMEBL3D/graph.json","fetch_events":"https://pith.science/api/pith-number/KA7ICOTDHZL4TITMUSLHMEBL3D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D/action/storage_attestation","attest_author":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D/action/author_attestation","sign_citation":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D/action/citation_signature","submit_replication":"https://pith.science/pith/KA7ICOTDHZL4TITMUSLHMEBL3D/action/replication_record"}},"created_at":"2026-05-26T01:03:21.257084+00:00","updated_at":"2026-05-26T01:03:21.257084+00:00"}