{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YXZ4JDPRXCLMU4MXAKZDBTIOLE","short_pith_number":"pith:YXZ4JDPR","schema_version":"1.0","canonical_sha256":"c5f3c48df1b896ca719702b230cd0e593a3bb436da19ba9382de86bc3765eef9","source":{"kind":"arxiv","id":"2508.06296","version":2},"attestation_state":"computed","paper":{"title":"LLM Robustness Leaderboard v1 --Technical report","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Nicolas Miailhe, Pierre Peign\\'e - Lefebvre, Quentin Feuillade-Montixi, Tom David","submitted_at":"2025-08-08T13:15:40Z","abstract_excerpt":"This technical report accompanies the LLM robustness leaderboard published by PRISM Eval for the Paris AI Action Summit. We introduce PRISM Eval Behavior Elicitation Tool (BET), an AI system performing automated red-teaming through Dynamic Adversarial Optimization that achieves 100% Attack Success Rate (ASR) against 37 of 41 state-of-the-art LLMs. Beyond binary success metrics, we propose a fine-grained robustness metric estimating the average number of attempts required to elicit harmful behaviors, revealing that attack difficulty varies by over 300-fold across models despite universal vulner"},"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":"2508.06296","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-08T13:15:40Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"28bf7510dd445e4749cdc8f6ec7777cd42cdd5b14c78932f7978f579f38629cd","abstract_canon_sha256":"90cbf67f8b608769929aa15a7569f80119508d519f8611ef1958ab024a28fa95"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:53:07.194678Z","signature_b64":"zUygfwFfiQ5zfGtEvMcr0wLGn50cL/s0182P1mtIBuT3nNC+TGt0BtZGE5DatHKNnMIx29fysFXnyISLpOIXAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c5f3c48df1b896ca719702b230cd0e593a3bb436da19ba9382de86bc3765eef9","last_reissued_at":"2026-07-05T11:53:07.194175Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:53:07.194175Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLM Robustness Leaderboard v1 --Technical report","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Nicolas Miailhe, Pierre Peign\\'e - Lefebvre, Quentin Feuillade-Montixi, Tom David","submitted_at":"2025-08-08T13:15:40Z","abstract_excerpt":"This technical report accompanies the LLM robustness leaderboard published by PRISM Eval for the Paris AI Action Summit. We introduce PRISM Eval Behavior Elicitation Tool (BET), an AI system performing automated red-teaming through Dynamic Adversarial Optimization that achieves 100% Attack Success Rate (ASR) against 37 of 41 state-of-the-art LLMs. Beyond binary success metrics, we propose a fine-grained robustness metric estimating the average number of attempts required to elicit harmful behaviors, revealing that attack difficulty varies by over 300-fold across models despite universal vulner"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.06296","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/2508.06296/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":"2508.06296","created_at":"2026-07-05T11:53:07.194242+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.06296v2","created_at":"2026-07-05T11:53:07.194242+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.06296","created_at":"2026-07-05T11:53:07.194242+00:00"},{"alias_kind":"pith_short_12","alias_value":"YXZ4JDPRXCLM","created_at":"2026-07-05T11:53:07.194242+00:00"},{"alias_kind":"pith_short_16","alias_value":"YXZ4JDPRXCLMU4MX","created_at":"2026-07-05T11:53:07.194242+00:00"},{"alias_kind":"pith_short_8","alias_value":"YXZ4JDPR","created_at":"2026-07-05T11:53:07.194242+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/YXZ4JDPRXCLMU4MXAKZDBTIOLE","json":"https://pith.science/pith/YXZ4JDPRXCLMU4MXAKZDBTIOLE.json","graph_json":"https://pith.science/api/pith-number/YXZ4JDPRXCLMU4MXAKZDBTIOLE/graph.json","events_json":"https://pith.science/api/pith-number/YXZ4JDPRXCLMU4MXAKZDBTIOLE/events.json","paper":"https://pith.science/paper/YXZ4JDPR"},"agent_actions":{"view_html":"https://pith.science/pith/YXZ4JDPRXCLMU4MXAKZDBTIOLE","download_json":"https://pith.science/pith/YXZ4JDPRXCLMU4MXAKZDBTIOLE.json","view_paper":"https://pith.science/paper/YXZ4JDPR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.06296&json=true","fetch_graph":"https://pith.science/api/pith-number/YXZ4JDPRXCLMU4MXAKZDBTIOLE/graph.json","fetch_events":"https://pith.science/api/pith-number/YXZ4JDPRXCLMU4MXAKZDBTIOLE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YXZ4JDPRXCLMU4MXAKZDBTIOLE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YXZ4JDPRXCLMU4MXAKZDBTIOLE/action/storage_attestation","attest_author":"https://pith.science/pith/YXZ4JDPRXCLMU4MXAKZDBTIOLE/action/author_attestation","sign_citation":"https://pith.science/pith/YXZ4JDPRXCLMU4MXAKZDBTIOLE/action/citation_signature","submit_replication":"https://pith.science/pith/YXZ4JDPRXCLMU4MXAKZDBTIOLE/action/replication_record"}},"created_at":"2026-07-05T11:53:07.194242+00:00","updated_at":"2026-07-05T11:53:07.194242+00:00"}