{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:T7OVY5SGELIXCO5MHXK72OXIOX","short_pith_number":"pith:T7OVY5SG","schema_version":"1.0","canonical_sha256":"9fdd5c764622d1713bac3dd5fd3ae875d018c03fe61877d3c81d5b5a67483159","source":{"kind":"arxiv","id":"2410.02440","version":2},"attestation_state":"computed","paper":{"title":"Optimizing Adaptive Attacks against Watermarks for Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Abdulrahman Diaa, Nils Lukas, Toluwani Aremu","submitted_at":"2024-10-03T12:37:39Z","abstract_excerpt":"Large Language Models (LLMs) can be misused to spread unwanted content at scale. Content watermarking deters misuse by hiding messages in content, enabling its detection using a secret watermarking key. Robustness is a core security property, stating that evading detection requires (significant) degradation of the content's quality. Many LLM watermarking methods have been proposed, but robustness is tested only against non-adaptive attackers who lack knowledge of the watermarking method and can find only suboptimal attacks. We formulate watermark robustness as an objective function and use pre"},"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":"2410.02440","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-10-03T12:37:39Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"b04900af2aceac3d6673e3581fc1706bb15b6f56da5b131f5b80f99b875a4f97","abstract_canon_sha256":"85d1c0a47ec63d3f34c0a9a2f7aca2a4b286cb212f168b17096ec95b74f869a5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:06:23.829814Z","signature_b64":"i6+a8zLxzqUIdUo76sTFr3khKGcAn5QXQP32OOEjOfqvQusq4j/TsWzakieUFLhENuPgPuZjnUuDABzq3+z0DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9fdd5c764622d1713bac3dd5fd3ae875d018c03fe61877d3c81d5b5a67483159","last_reissued_at":"2026-07-05T11:06:23.829249Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:06:23.829249Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Optimizing Adaptive Attacks against Watermarks for Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Abdulrahman Diaa, Nils Lukas, Toluwani Aremu","submitted_at":"2024-10-03T12:37:39Z","abstract_excerpt":"Large Language Models (LLMs) can be misused to spread unwanted content at scale. Content watermarking deters misuse by hiding messages in content, enabling its detection using a secret watermarking key. Robustness is a core security property, stating that evading detection requires (significant) degradation of the content's quality. Many LLM watermarking methods have been proposed, but robustness is tested only against non-adaptive attackers who lack knowledge of the watermarking method and can find only suboptimal attacks. We formulate watermark robustness as an objective function and use pre"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.02440","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/2410.02440/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":"2410.02440","created_at":"2026-07-05T11:06:23.829298+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.02440v2","created_at":"2026-07-05T11:06:23.829298+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.02440","created_at":"2026-07-05T11:06:23.829298+00:00"},{"alias_kind":"pith_short_12","alias_value":"T7OVY5SGELIX","created_at":"2026-07-05T11:06:23.829298+00:00"},{"alias_kind":"pith_short_16","alias_value":"T7OVY5SGELIXCO5M","created_at":"2026-07-05T11:06:23.829298+00:00"},{"alias_kind":"pith_short_8","alias_value":"T7OVY5SG","created_at":"2026-07-05T11:06:23.829298+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.07871","citing_title":"Mitigating Watermark Forgery in Generative Models via Randomized Key Selection","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2510.18333","citing_title":"Position: LLM Watermarking Should Align Stakeholders' Incentives for Practical Adoption","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11546","citing_title":"RLSpoofer: A Lightweight Evaluator for LLM Watermark Spoofing Resilience","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12216","citing_title":"TimeMark: A Trustworthy Time Watermarking Framework for Exact Generation-Time Recovery from AIGC","ref_index":13,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX","json":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX.json","graph_json":"https://pith.science/api/pith-number/T7OVY5SGELIXCO5MHXK72OXIOX/graph.json","events_json":"https://pith.science/api/pith-number/T7OVY5SGELIXCO5MHXK72OXIOX/events.json","paper":"https://pith.science/paper/T7OVY5SG"},"agent_actions":{"view_html":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX","download_json":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX.json","view_paper":"https://pith.science/paper/T7OVY5SG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.02440&json=true","fetch_graph":"https://pith.science/api/pith-number/T7OVY5SGELIXCO5MHXK72OXIOX/graph.json","fetch_events":"https://pith.science/api/pith-number/T7OVY5SGELIXCO5MHXK72OXIOX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX/action/storage_attestation","attest_author":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX/action/author_attestation","sign_citation":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX/action/citation_signature","submit_replication":"https://pith.science/pith/T7OVY5SGELIXCO5MHXK72OXIOX/action/replication_record"}},"created_at":"2026-07-05T11:06:23.829298+00:00","updated_at":"2026-07-05T11:06:23.829298+00:00"}