{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:AJZUOYF24LQPEPTHH64EFM5ELE","short_pith_number":"pith:AJZUOYF2","schema_version":"1.0","canonical_sha256":"02734760bae2e0f23e673fb842b3a4593ed4e2ed8064034991dfef75ca7bdea5","source":{"kind":"arxiv","id":"2607.11070","version":1},"attestation_state":"computed","paper":{"title":"MJ: Multi-turn LLM Jailbreaking via Decomposed Credit Assignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jihoon Cho, Junyoung Park, Namgyu Park, Sangdon Park, Sechan Lee, Yoon-Chan Jhi","submitted_at":"2026-07-13T04:19:37Z","abstract_excerpt":"Modern large language models (LLMs) operate in interactive multi-turn settings, making multi-turn jailbreaking a realistic threat model and an important setting for automated red teaming. A core challenge in learning multi-turn jailbreak attackers is credit assignment: different turns contribute differently to the final outcome, yet existing learning signals are often too coarse to identify their individual contributions. We propose decomposed credit GRPO (DC-GRPO), a unified turn-level credit assignment framework for Group Relative Policy Optimization in multi-turn jailbreak learning. DC-GRPO"},"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":"2607.11070","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-07-13T04:19:37Z","cross_cats_sorted":[],"title_canon_sha256":"a188a9d6d46139f4616ff21a421ccfe8c4dfe30d15ad5bc932358cf29505f478","abstract_canon_sha256":"e76fdaa55c02ed6f303fb28e50a64773860b70573010ae58c67ca4b82027e76a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-14T01:22:14.614793Z","signature_b64":"Z1LCmNXuR4dNGDWYsxCilC4lSEE3aP9vUmqAmRzfeCqh9WfqMWvAS7Pu0aF8QAWTCHCbJRHpCLEZyH+eyfO+BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"02734760bae2e0f23e673fb842b3a4593ed4e2ed8064034991dfef75ca7bdea5","last_reissued_at":"2026-07-14T01:22:14.613904Z","signature_status":"signed_v1","first_computed_at":"2026-07-14T01:22:14.613904Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MJ: Multi-turn LLM Jailbreaking via Decomposed Credit Assignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jihoon Cho, Junyoung Park, Namgyu Park, Sangdon Park, Sechan Lee, Yoon-Chan Jhi","submitted_at":"2026-07-13T04:19:37Z","abstract_excerpt":"Modern large language models (LLMs) operate in interactive multi-turn settings, making multi-turn jailbreaking a realistic threat model and an important setting for automated red teaming. A core challenge in learning multi-turn jailbreak attackers is credit assignment: different turns contribute differently to the final outcome, yet existing learning signals are often too coarse to identify their individual contributions. We propose decomposed credit GRPO (DC-GRPO), a unified turn-level credit assignment framework for Group Relative Policy Optimization in multi-turn jailbreak learning. DC-GRPO"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.11070","kind":"arxiv","version":1},"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/2607.11070/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":"2607.11070","created_at":"2026-07-14T01:22:14.614369+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.11070v1","created_at":"2026-07-14T01:22:14.614369+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.11070","created_at":"2026-07-14T01:22:14.614369+00:00"},{"alias_kind":"pith_short_12","alias_value":"AJZUOYF24LQP","created_at":"2026-07-14T01:22:14.614369+00:00"},{"alias_kind":"pith_short_16","alias_value":"AJZUOYF24LQPEPTH","created_at":"2026-07-14T01:22:14.614369+00:00"},{"alias_kind":"pith_short_8","alias_value":"AJZUOYF2","created_at":"2026-07-14T01:22:14.614369+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/AJZUOYF24LQPEPTHH64EFM5ELE","json":"https://pith.science/pith/AJZUOYF24LQPEPTHH64EFM5ELE.json","graph_json":"https://pith.science/api/pith-number/AJZUOYF24LQPEPTHH64EFM5ELE/graph.json","events_json":"https://pith.science/api/pith-number/AJZUOYF24LQPEPTHH64EFM5ELE/events.json","paper":"https://pith.science/paper/AJZUOYF2"},"agent_actions":{"view_html":"https://pith.science/pith/AJZUOYF24LQPEPTHH64EFM5ELE","download_json":"https://pith.science/pith/AJZUOYF24LQPEPTHH64EFM5ELE.json","view_paper":"https://pith.science/paper/AJZUOYF2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.11070&json=true","fetch_graph":"https://pith.science/api/pith-number/AJZUOYF24LQPEPTHH64EFM5ELE/graph.json","fetch_events":"https://pith.science/api/pith-number/AJZUOYF24LQPEPTHH64EFM5ELE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AJZUOYF24LQPEPTHH64EFM5ELE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AJZUOYF24LQPEPTHH64EFM5ELE/action/storage_attestation","attest_author":"https://pith.science/pith/AJZUOYF24LQPEPTHH64EFM5ELE/action/author_attestation","sign_citation":"https://pith.science/pith/AJZUOYF24LQPEPTHH64EFM5ELE/action/citation_signature","submit_replication":"https://pith.science/pith/AJZUOYF24LQPEPTHH64EFM5ELE/action/replication_record"}},"created_at":"2026-07-14T01:22:14.614369+00:00","updated_at":"2026-07-14T01:22:14.614369+00:00"}