{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:TJZ73K56GALY5WZ5XIPMPMB6KG","short_pith_number":"pith:TJZ73K56","canonical_record":{"source":{"id":"2605.20286","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-05-19T06:11:04Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"d4648fb7240e0a6f72fcb76f773b3855850bfe447562a27ad2bc46e83ec578ca","abstract_canon_sha256":"836fb5368136ba72f02db7dd3476bbaca6249816cc0ff40ef204db8e3a124f1b"},"schema_version":"1.0"},"canonical_sha256":"9a73fdabbe30178edb3dba1ec7b03e51a6f52777a6ee16258e66025d1b1f7243","source":{"kind":"arxiv","id":"2605.20286","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.20286","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"arxiv_version","alias_value":"2605.20286v1","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.20286","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"pith_short_12","alias_value":"TJZ73K56GALY","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"pith_short_16","alias_value":"TJZ73K56GALY5WZ5","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"pith_short_8","alias_value":"TJZ73K56","created_at":"2026-05-21T00:04:24Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:TJZ73K56GALY5WZ5XIPMPMB6KG","target":"record","payload":{"canonical_record":{"source":{"id":"2605.20286","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-05-19T06:11:04Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"d4648fb7240e0a6f72fcb76f773b3855850bfe447562a27ad2bc46e83ec578ca","abstract_canon_sha256":"836fb5368136ba72f02db7dd3476bbaca6249816cc0ff40ef204db8e3a124f1b"},"schema_version":"1.0"},"canonical_sha256":"9a73fdabbe30178edb3dba1ec7b03e51a6f52777a6ee16258e66025d1b1f7243","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T00:04:24.026370Z","signature_b64":"FqRSvN3HK1XZjCeMCAobVJibJG6u9wDQd6aETuWCtaCiZVoxzoR9+4REKhKlmMRBIh9mQK/ipKu5xpd4cb1TAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9a73fdabbe30178edb3dba1ec7b03e51a6f52777a6ee16258e66025d1b1f7243","last_reissued_at":"2026-05-21T00:04:24.025843Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T00:04:24.025843Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.20286","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-21T00:04:24Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"KivQxYvcKbmfK0TTT2U6gPqwc6xYP3nXlQn+czz57E7VSFgrh4zfKZ7DCkTYJhq74Y9z2Yj9Xkb1O0zAU+67Aw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T21:46:33.027967Z"},"content_sha256":"08f6b12c8394930af31e298da65264cb202e556849aca38b12ee8c45fde671ec","schema_version":"1.0","event_id":"sha256:08f6b12c8394930af31e298da65264cb202e556849aca38b12ee8c45fde671ec"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:TJZ73K56GALY5WZ5XIPMPMB6KG","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Adaptive Probe-based Steering for Robust LLM Jailbreaking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Junhao Dong, Junxi Chen, Xiaohua Xie","submitted_at":"2026-05-19T06:11:04Z","abstract_excerpt":"Recent work has demonstrated the potential of contrastive steering for jailbreaking Large Language Models (LLMs). However, existing methods rely on limited and inherently biased contrastive prompts and require laborious manual tuning of steering strength, limiting their robustness and effectiveness. In this paper, we leverage the idea of model extraction to guide the learned steering vectors to approximate the ideal one and propose tuning the steering strength adaptively based on contrastive activations' statistics. Experiments demonstrate that our method notably improves the effectiveness and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.20286","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/2605.20286/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":"2026-05-21T00:04:24Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1CyxaO3J3z06juCJ2mGx5KRyYbYMkXXirboXYnLXZ38/KeDbkuH1j82O5fVY0MiY347FfEoJhrSVacIeTOYcDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T21:46:33.028397Z"},"content_sha256":"496a4bdeb74fdc394b929eefefb98ddeaefaf6d2b952322639a0f51a74157788","schema_version":"1.0","event_id":"sha256:496a4bdeb74fdc394b929eefefb98ddeaefaf6d2b952322639a0f51a74157788"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TJZ73K56GALY5WZ5XIPMPMB6KG/bundle.json","state_url":"https://pith.science/pith/TJZ73K56GALY5WZ5XIPMPMB6KG/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TJZ73K56GALY5WZ5XIPMPMB6KG/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":"2026-05-25T21:46:33Z","links":{"resolver":"https://pith.science/pith/TJZ73K56GALY5WZ5XIPMPMB6KG","bundle":"https://pith.science/pith/TJZ73K56GALY5WZ5XIPMPMB6KG/bundle.json","state":"https://pith.science/pith/TJZ73K56GALY5WZ5XIPMPMB6KG/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TJZ73K56GALY5WZ5XIPMPMB6KG/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:TJZ73K56GALY5WZ5XIPMPMB6KG","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":"836fb5368136ba72f02db7dd3476bbaca6249816cc0ff40ef204db8e3a124f1b","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-05-19T06:11:04Z","title_canon_sha256":"d4648fb7240e0a6f72fcb76f773b3855850bfe447562a27ad2bc46e83ec578ca"},"schema_version":"1.0","source":{"id":"2605.20286","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.20286","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"arxiv_version","alias_value":"2605.20286v1","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.20286","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"pith_short_12","alias_value":"TJZ73K56GALY","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"pith_short_16","alias_value":"TJZ73K56GALY5WZ5","created_at":"2026-05-21T00:04:24Z"},{"alias_kind":"pith_short_8","alias_value":"TJZ73K56","created_at":"2026-05-21T00:04:24Z"}],"graph_snapshots":[{"event_id":"sha256:496a4bdeb74fdc394b929eefefb98ddeaefaf6d2b952322639a0f51a74157788","target":"graph","created_at":"2026-05-21T00:04:24Z","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/2605.20286/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent work has demonstrated the potential of contrastive steering for jailbreaking Large Language Models (LLMs). However, existing methods rely on limited and inherently biased contrastive prompts and require laborious manual tuning of steering strength, limiting their robustness and effectiveness. In this paper, we leverage the idea of model extraction to guide the learned steering vectors to approximate the ideal one and propose tuning the steering strength adaptively based on contrastive activations' statistics. Experiments demonstrate that our method notably improves the effectiveness and","authors_text":"Junhao Dong, Junxi Chen, Xiaohua Xie","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-05-19T06:11:04Z","title":"Adaptive Probe-based Steering for Robust LLM Jailbreaking"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.20286","kind":"arxiv","version":1},"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":"sha256:08f6b12c8394930af31e298da65264cb202e556849aca38b12ee8c45fde671ec","target":"record","created_at":"2026-05-21T00:04:24Z","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":"836fb5368136ba72f02db7dd3476bbaca6249816cc0ff40ef204db8e3a124f1b","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-05-19T06:11:04Z","title_canon_sha256":"d4648fb7240e0a6f72fcb76f773b3855850bfe447562a27ad2bc46e83ec578ca"},"schema_version":"1.0","source":{"id":"2605.20286","kind":"arxiv","version":1}},"canonical_sha256":"9a73fdabbe30178edb3dba1ec7b03e51a6f52777a6ee16258e66025d1b1f7243","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9a73fdabbe30178edb3dba1ec7b03e51a6f52777a6ee16258e66025d1b1f7243","first_computed_at":"2026-05-21T00:04:24.025843Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-21T00:04:24.025843Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"FqRSvN3HK1XZjCeMCAobVJibJG6u9wDQd6aETuWCtaCiZVoxzoR9+4REKhKlmMRBIh9mQK/ipKu5xpd4cb1TAA==","signature_status":"signed_v1","signed_at":"2026-05-21T00:04:24.026370Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.20286","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:08f6b12c8394930af31e298da65264cb202e556849aca38b12ee8c45fde671ec","sha256:496a4bdeb74fdc394b929eefefb98ddeaefaf6d2b952322639a0f51a74157788"],"state_sha256":"39a1f130ef748aeabc7fb0e7ddf939402227fa4bb9ca7265f0afd66510014704"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Hse4ylo0m4bi7zeLnTtl47h/vzZGC2ZP5Om+h4gDiwXp5MhDE6YERVv1LRydAI983/VZwzk0HueLBmTtKSORBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-25T21:46:33.030980Z","bundle_sha256":"e16c0f12b9d768712da1f941df9503b418355a5a0cab6da5a7a223a77025b4cb"}}