{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:ZUMFG6NLCAPAX2KBICKUIN4UOM","short_pith_number":"pith:ZUMFG6NL","canonical_record":{"source":{"id":"2501.16727","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-01-28T06:07:58Z","cross_cats_sorted":[],"title_canon_sha256":"8e05dad2884fc25453c3a7ddc67ccd15b51d3e98eb2ed643fc3f79751d0a8c76","abstract_canon_sha256":"f7201d78d5eefb8ae27d75b5818b6acdbc033001f522d27867e85b8693ef5789"},"schema_version":"1.0"},"canonical_sha256":"cd185379ab101e0be9414095443794732f79af7d3a7e63989e6fe67748a27180","source":{"kind":"arxiv","id":"2501.16727","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.16727","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"arxiv_version","alias_value":"2501.16727v2","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.16727","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"pith_short_12","alias_value":"ZUMFG6NLCAPA","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"pith_short_16","alias_value":"ZUMFG6NLCAPAX2KB","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"pith_short_8","alias_value":"ZUMFG6NL","created_at":"2026-07-05T10:07:29Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:ZUMFG6NLCAPAX2KBICKUIN4UOM","target":"record","payload":{"canonical_record":{"source":{"id":"2501.16727","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-01-28T06:07:58Z","cross_cats_sorted":[],"title_canon_sha256":"8e05dad2884fc25453c3a7ddc67ccd15b51d3e98eb2ed643fc3f79751d0a8c76","abstract_canon_sha256":"f7201d78d5eefb8ae27d75b5818b6acdbc033001f522d27867e85b8693ef5789"},"schema_version":"1.0"},"canonical_sha256":"cd185379ab101e0be9414095443794732f79af7d3a7e63989e6fe67748a27180","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:07:29.999109Z","signature_b64":"8qH7IOoc3KTonrGoKvFIzS4684ia/wn/a/ddoTPnWitHo/XLKN+nU6wvciimiPuEageQSp1F2odBUwVir/TbBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cd185379ab101e0be9414095443794732f79af7d3a7e63989e6fe67748a27180","last_reissued_at":"2026-07-05T10:07:29.998634Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:07:29.998634Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2501.16727","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":"2026-07-05T10:07:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"U/8IO76Bq5vP8xC29Zr9p5kWNtX3IDZYL1RjkhSzO/8hXTyiKL/JKihbFUmSM2Zqst0m0xvyJF9bcSxQc77qDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T05:45:13.652047Z"},"content_sha256":"7e96a2edfe4555e5d73ffec6094f8676f584fe8da9bc7c27f51284d8562bf812","schema_version":"1.0","event_id":"sha256:7e96a2edfe4555e5d73ffec6094f8676f584fe8da9bc7c27f51284d8562bf812"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:ZUMFG6NLCAPAX2KBICKUIN4UOM","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"xJailbreak: Representation Space Guided Reinforcement Learning for Interpretable LLM Jailbreaking","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ahmadreza Argha, Chi Wei, Hamid Alinejad-Rokny, Liyang Fan, Min Yang, Ruifeng Xu, Shiwen Ni, Shuaimin Li, Sunbowen Lee, Yicheng Gong","submitted_at":"2025-01-28T06:07:58Z","abstract_excerpt":"Safety alignment mechanism are essential for preventing large language models (LLMs) from generating harmful information or unethical content. However, cleverly crafted prompts can bypass these safety measures without accessing the model's internal parameters, a phenomenon known as black-box jailbreak. Existing heuristic black-box attack methods, such as genetic algorithms, suffer from limited effectiveness due to their inherent randomness, while recent reinforcement learning (RL) based methods often lack robust and informative reward signals. To address these challenges, we propose a novel bl"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.16727","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/2501.16727/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":"2026-07-05T10:07:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Oah/lBfwx69FvUVaRfw7B36mqwF8yciCcuuTk9buHW+qoqZgKe4y7JctDz7r+gdpTsZTKtUGiLZcbSy0kaRDAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T05:45:13.652439Z"},"content_sha256":"c0560d288ae534d4608ac80f83ce1ff2463022a9cee5e71e9991b6227e220960","schema_version":"1.0","event_id":"sha256:c0560d288ae534d4608ac80f83ce1ff2463022a9cee5e71e9991b6227e220960"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ZUMFG6NLCAPAX2KBICKUIN4UOM/bundle.json","state_url":"https://pith.science/pith/ZUMFG6NLCAPAX2KBICKUIN4UOM/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ZUMFG6NLCAPAX2KBICKUIN4UOM/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":"2026-07-07T05:45:13Z","links":{"resolver":"https://pith.science/pith/ZUMFG6NLCAPAX2KBICKUIN4UOM","bundle":"https://pith.science/pith/ZUMFG6NLCAPAX2KBICKUIN4UOM/bundle.json","state":"https://pith.science/pith/ZUMFG6NLCAPAX2KBICKUIN4UOM/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ZUMFG6NLCAPAX2KBICKUIN4UOM/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:ZUMFG6NLCAPAX2KBICKUIN4UOM","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":"f7201d78d5eefb8ae27d75b5818b6acdbc033001f522d27867e85b8693ef5789","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-01-28T06:07:58Z","title_canon_sha256":"8e05dad2884fc25453c3a7ddc67ccd15b51d3e98eb2ed643fc3f79751d0a8c76"},"schema_version":"1.0","source":{"id":"2501.16727","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.16727","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"arxiv_version","alias_value":"2501.16727v2","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.16727","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"pith_short_12","alias_value":"ZUMFG6NLCAPA","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"pith_short_16","alias_value":"ZUMFG6NLCAPAX2KB","created_at":"2026-07-05T10:07:29Z"},{"alias_kind":"pith_short_8","alias_value":"ZUMFG6NL","created_at":"2026-07-05T10:07:29Z"}],"graph_snapshots":[{"event_id":"sha256:c0560d288ae534d4608ac80f83ce1ff2463022a9cee5e71e9991b6227e220960","target":"graph","created_at":"2026-07-05T10:07:29Z","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/2501.16727/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Safety alignment mechanism are essential for preventing large language models (LLMs) from generating harmful information or unethical content. However, cleverly crafted prompts can bypass these safety measures without accessing the model's internal parameters, a phenomenon known as black-box jailbreak. Existing heuristic black-box attack methods, such as genetic algorithms, suffer from limited effectiveness due to their inherent randomness, while recent reinforcement learning (RL) based methods often lack robust and informative reward signals. To address these challenges, we propose a novel bl","authors_text":"Ahmadreza Argha, Chi Wei, Hamid Alinejad-Rokny, Liyang Fan, Min Yang, Ruifeng Xu, Shiwen Ni, Shuaimin Li, Sunbowen Lee, Yicheng Gong","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-01-28T06:07:58Z","title":"xJailbreak: Representation Space Guided Reinforcement Learning for Interpretable LLM Jailbreaking"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.16727","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":"sha256:7e96a2edfe4555e5d73ffec6094f8676f584fe8da9bc7c27f51284d8562bf812","target":"record","created_at":"2026-07-05T10:07:29Z","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":"f7201d78d5eefb8ae27d75b5818b6acdbc033001f522d27867e85b8693ef5789","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-01-28T06:07:58Z","title_canon_sha256":"8e05dad2884fc25453c3a7ddc67ccd15b51d3e98eb2ed643fc3f79751d0a8c76"},"schema_version":"1.0","source":{"id":"2501.16727","kind":"arxiv","version":2}},"canonical_sha256":"cd185379ab101e0be9414095443794732f79af7d3a7e63989e6fe67748a27180","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"cd185379ab101e0be9414095443794732f79af7d3a7e63989e6fe67748a27180","first_computed_at":"2026-07-05T10:07:29.998634Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:07:29.998634Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"8qH7IOoc3KTonrGoKvFIzS4684ia/wn/a/ddoTPnWitHo/XLKN+nU6wvciimiPuEageQSp1F2odBUwVir/TbBA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:07:29.999109Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.16727","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7e96a2edfe4555e5d73ffec6094f8676f584fe8da9bc7c27f51284d8562bf812","sha256:c0560d288ae534d4608ac80f83ce1ff2463022a9cee5e71e9991b6227e220960"],"state_sha256":"702610ab1b8a581604232a4b732d76cc3cf03df0f3da04cb883bdbb6238210c5"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"e7ddgQnizcCwVLDaGU2/KKyI9ji4CEQgRkwK6HpTEDjwu8re/q/cJNhahBFM8mqDpw4rWZTRmSUdVnlFGUVCAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T05:45:13.655003Z","bundle_sha256":"869e0542d2809a6d829fa5563c3ab2d4a3549ca2d130ae69879c71bcfca2ea63"}}