{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:ENNDSNET4RK3PYNUS6LTY3KJMR","short_pith_number":"pith:ENNDSNET","canonical_record":{"source":{"id":"2503.06269","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-08T16:29:45Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"17198aea37d3de4d9afab909bdab860c77d34dc53010e5296c3d298be823a46e","abstract_canon_sha256":"927c5e58856edf08afb7b7a238f412718c0192f9c56a023d9433c1ffc725824f"},"schema_version":"1.0"},"canonical_sha256":"235a393493e455b7e1b497973c6d496455e16fca0d3b13676cf5f84502e74dde","source":{"kind":"arxiv","id":"2503.06269","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.06269","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"arxiv_version","alias_value":"2503.06269v2","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.06269","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"pith_short_12","alias_value":"ENNDSNET4RK3","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"pith_short_16","alias_value":"ENNDSNET4RK3PYNU","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"pith_short_8","alias_value":"ENNDSNET","created_at":"2026-07-05T10:59:06Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:ENNDSNET4RK3PYNUS6LTY3KJMR","target":"record","payload":{"canonical_record":{"source":{"id":"2503.06269","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-08T16:29:45Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"17198aea37d3de4d9afab909bdab860c77d34dc53010e5296c3d298be823a46e","abstract_canon_sha256":"927c5e58856edf08afb7b7a238f412718c0192f9c56a023d9433c1ffc725824f"},"schema_version":"1.0"},"canonical_sha256":"235a393493e455b7e1b497973c6d496455e16fca0d3b13676cf5f84502e74dde","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:59:06.807915Z","signature_b64":"oeQ7WexCe2chxwEVM4qC81q6sumLI9IFg6HGroWF7AviC2GiMXJRTj6Cse6GMyZnHkgxbKzHzwuD4/pndzO4Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"235a393493e455b7e1b497973c6d496455e16fca0d3b13676cf5f84502e74dde","last_reissued_at":"2026-07-05T10:59:06.807402Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:59:06.807402Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2503.06269","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:59:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"gALR21pwSlhNJ329cx0ndBNxLGtsHPuE8AOpj6XCprIRqqnX9sGroKmhtJ5E8t5O0h0tudw4c1pV+XhH5delBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T20:33:32.102878Z"},"content_sha256":"c87a4b24c86678a714d36cf4b413eff83ea93e77cd18ac09e3236801a5fa9cf0","schema_version":"1.0","event_id":"sha256:c87a4b24c86678a714d36cf4b413eff83ea93e77cd18ac09e3236801a5fa9cf0"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:ENNDSNET4RK3PYNUS6LTY3KJMR","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Using Mechanistic Interpretability to Craft Adversarial Attacks against Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Boussad Addad, Katarzyna Kapusta, Thomas Winninger","submitted_at":"2025-03-08T16:29:45Z","abstract_excerpt":"Traditional white-box methods for creating adversarial perturbations against LLMs typically rely only on gradient computation from the targeted model, ignoring the internal mechanisms responsible for attack success or failure. Conversely, interpretability studies that analyze these internal mechanisms lack practical applications beyond runtime interventions. We bridge this gap by introducing a novel white-box approach that leverages mechanistic interpretability techniques to craft practical adversarial inputs. Specifically, we first identify acceptance subspaces - sets of feature vectors that "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.06269","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/2503.06269/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:59:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"qM4o6zJA/PCXB11FQawCIEjU31kiHuoCXNXj1N7sK7jtlsayMqrvloG+eIyHouqYtC4PRWHjYowjJLB805BeBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T20:33:32.103818Z"},"content_sha256":"0325a72f48e40ce047a58bc3e82f6019aa92394393b1a3eaafc3fb0ec37b2c82","schema_version":"1.0","event_id":"sha256:0325a72f48e40ce047a58bc3e82f6019aa92394393b1a3eaafc3fb0ec37b2c82"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ENNDSNET4RK3PYNUS6LTY3KJMR/bundle.json","state_url":"https://pith.science/pith/ENNDSNET4RK3PYNUS6LTY3KJMR/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ENNDSNET4RK3PYNUS6LTY3KJMR/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":"2026-08-07T20:33:32Z","links":{"resolver":"https://pith.science/pith/ENNDSNET4RK3PYNUS6LTY3KJMR","bundle":"https://pith.science/pith/ENNDSNET4RK3PYNUS6LTY3KJMR/bundle.json","state":"https://pith.science/pith/ENNDSNET4RK3PYNUS6LTY3KJMR/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ENNDSNET4RK3PYNUS6LTY3KJMR/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:ENNDSNET4RK3PYNUS6LTY3KJMR","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":"927c5e58856edf08afb7b7a238f412718c0192f9c56a023d9433c1ffc725824f","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-08T16:29:45Z","title_canon_sha256":"17198aea37d3de4d9afab909bdab860c77d34dc53010e5296c3d298be823a46e"},"schema_version":"1.0","source":{"id":"2503.06269","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.06269","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"arxiv_version","alias_value":"2503.06269v2","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.06269","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"pith_short_12","alias_value":"ENNDSNET4RK3","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"pith_short_16","alias_value":"ENNDSNET4RK3PYNU","created_at":"2026-07-05T10:59:06Z"},{"alias_kind":"pith_short_8","alias_value":"ENNDSNET","created_at":"2026-07-05T10:59:06Z"}],"graph_snapshots":[{"event_id":"sha256:0325a72f48e40ce047a58bc3e82f6019aa92394393b1a3eaafc3fb0ec37b2c82","target":"graph","created_at":"2026-07-05T10:59:06Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2503.06269/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Traditional white-box methods for creating adversarial perturbations against LLMs typically rely only on gradient computation from the targeted model, ignoring the internal mechanisms responsible for attack success or failure. Conversely, interpretability studies that analyze these internal mechanisms lack practical applications beyond runtime interventions. We bridge this gap by introducing a novel white-box approach that leverages mechanistic interpretability techniques to craft practical adversarial inputs. Specifically, we first identify acceptance subspaces - sets of feature vectors that ","authors_text":"Boussad Addad, Katarzyna Kapusta, Thomas Winninger","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-08T16:29:45Z","title":"Using Mechanistic Interpretability to Craft Adversarial Attacks against Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.06269","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c87a4b24c86678a714d36cf4b413eff83ea93e77cd18ac09e3236801a5fa9cf0","target":"record","created_at":"2026-07-05T10:59:06Z","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":"927c5e58856edf08afb7b7a238f412718c0192f9c56a023d9433c1ffc725824f","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-08T16:29:45Z","title_canon_sha256":"17198aea37d3de4d9afab909bdab860c77d34dc53010e5296c3d298be823a46e"},"schema_version":"1.0","source":{"id":"2503.06269","kind":"arxiv","version":2}},"canonical_sha256":"235a393493e455b7e1b497973c6d496455e16fca0d3b13676cf5f84502e74dde","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"235a393493e455b7e1b497973c6d496455e16fca0d3b13676cf5f84502e74dde","first_computed_at":"2026-07-05T10:59:06.807402Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:59:06.807402Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"oeQ7WexCe2chxwEVM4qC81q6sumLI9IFg6HGroWF7AviC2GiMXJRTj6Cse6GMyZnHkgxbKzHzwuD4/pndzO4Ag==","signature_status":"signed_v1","signed_at":"2026-07-05T10:59:06.807915Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.06269","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c87a4b24c86678a714d36cf4b413eff83ea93e77cd18ac09e3236801a5fa9cf0","sha256:0325a72f48e40ce047a58bc3e82f6019aa92394393b1a3eaafc3fb0ec37b2c82"],"state_sha256":"b649eeb84b2a5ecbb2744fab33ca1af9ba651733736a3fedd1e85525d3110b2e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"nnyToNot9rO6MrqfYS9woV3GAVL0/QEQxRNjtICiO3QiOrbLQfOjLmju00CHkHin1kOadsTfnTyxhHPn/yv1AA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T20:33:32.109445Z","bundle_sha256":"ae51bd04af3a9e93d706bd1789abdcdc4544a488a9bafa32ea361e3ebc1e0e9f"}}