{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:GU46XNRFSZ5QGHXYGS2LDZDINY","short_pith_number":"pith:GU46XNRF","canonical_record":{"source":{"id":"2608.09624","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-08-10T14:05:32Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"92b47b7bcfa958687ee20ca041649dda96012208251b9503d239fa1a02cb6945","abstract_canon_sha256":"788432c62332bf7dcdff48b971d36d1b43298771b265f4cb57878ab772149e78"},"schema_version":"1.0"},"canonical_sha256":"3539ebb625967b031ef834b4b1e4686e0d3f49fb35755ae837ae6f1dd2e94eb4","source":{"kind":"arxiv","id":"2608.09624","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.09624","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"arxiv_version","alias_value":"2608.09624v1","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.09624","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"pith_short_12","alias_value":"GU46XNRFSZ5Q","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"pith_short_16","alias_value":"GU46XNRFSZ5QGHXY","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"pith_short_8","alias_value":"GU46XNRF","created_at":"2026-08-11T02:24:38Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:GU46XNRFSZ5QGHXYGS2LDZDINY","target":"record","payload":{"canonical_record":{"source":{"id":"2608.09624","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-08-10T14:05:32Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"92b47b7bcfa958687ee20ca041649dda96012208251b9503d239fa1a02cb6945","abstract_canon_sha256":"788432c62332bf7dcdff48b971d36d1b43298771b265f4cb57878ab772149e78"},"schema_version":"1.0"},"canonical_sha256":"3539ebb625967b031ef834b4b1e4686e0d3f49fb35755ae837ae6f1dd2e94eb4","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T02:24:38.671488Z","signature_b64":"eewHBNHD8ZOFTcvQTJFpYD4AQyS30wLl6DPJKuH0zhGtzpB3Q52QHppCvcFprjAjXrI9PxWU9dqQ61I33Z9sCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3539ebb625967b031ef834b4b1e4686e0d3f49fb35755ae837ae6f1dd2e94eb4","last_reissued_at":"2026-08-11T02:24:38.669806Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T02:24:38.669806Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2608.09624","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":"2026-08-11T02:24:38Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Mcfmj2OhUw+P8yNUp3F2Z8Dq0p/wH829l1Pffls//4KJXKYMIxnIvcjuYFa+xBNOk/wiE1kAHYg32IpyqybTCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T22:19:55.789426Z"},"content_sha256":"1e66439aafaea600acadda08bd462a71d757280490214081a0ace91a234ef580","schema_version":"1.0","event_id":"sha256:1e66439aafaea600acadda08bd462a71d757280490214081a0ace91a234ef580"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:GU46XNRFSZ5QGHXYGS2LDZDINY","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Measuring the Wrong Thing: Internal Harmfulness Scores Anti-Rank Successful Jailbreaks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR"],"primary_cat":"cs.CL","authors_text":"Ci Tao, Jun Dai, Ming Deng, Mingyu Luo, Ping Chen, Sijin Sun, Xiaoyan Sun, Xue Tan, Yangfu Li, Yiming Cheng, Zilang Qiu","submitted_at":"2026-08-10T14:05:32Z","abstract_excerpt":"Internal safety scores judge a prompt before any text is generated, and they are validated by how well they separate harmful prompts from benign ones. That separation is then read as evidence that the score will also catch the attacks that succeed. Harmful intent is a property of the prompt. Jailbreak success is an outcome produced later by a particular target model, decoding policy, and judge. A filter tuned on a score that measures the wrong quantity spends its false positive budget on attacks that would have failed anyway. In this paper we audit that inference. Attention based measurements "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.09624","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/2608.09624/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":"2026-08-11T02:24:38Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HPGKMJy4ZDDb6r02qeZrIIY5hU9neuPuGeG4Pa+RyWBVvvct4s230Gl/GEdTI3oHNJaHZEbR4jAdQM7BkLc9Cw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T22:19:55.789944Z"},"content_sha256":"92728fcaaa5d1a503501df73da7acb9417cdf90e5eb3ad21ae313c3762516147","schema_version":"1.0","event_id":"sha256:92728fcaaa5d1a503501df73da7acb9417cdf90e5eb3ad21ae313c3762516147"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/GU46XNRFSZ5QGHXYGS2LDZDINY/bundle.json","state_url":"https://pith.science/pith/GU46XNRFSZ5QGHXYGS2LDZDINY/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/GU46XNRFSZ5QGHXYGS2LDZDINY/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":"2026-08-12T22:19:55Z","links":{"resolver":"https://pith.science/pith/GU46XNRFSZ5QGHXYGS2LDZDINY","bundle":"https://pith.science/pith/GU46XNRFSZ5QGHXYGS2LDZDINY/bundle.json","state":"https://pith.science/pith/GU46XNRFSZ5QGHXYGS2LDZDINY/state.json","well_known_bundle":"https://pith.science/.well-known/pith/GU46XNRFSZ5QGHXYGS2LDZDINY/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:GU46XNRFSZ5QGHXYGS2LDZDINY","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":"788432c62332bf7dcdff48b971d36d1b43298771b265f4cb57878ab772149e78","cross_cats_sorted":["cs.AI","cs.CR"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-08-10T14:05:32Z","title_canon_sha256":"92b47b7bcfa958687ee20ca041649dda96012208251b9503d239fa1a02cb6945"},"schema_version":"1.0","source":{"id":"2608.09624","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.09624","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"arxiv_version","alias_value":"2608.09624v1","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.09624","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"pith_short_12","alias_value":"GU46XNRFSZ5Q","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"pith_short_16","alias_value":"GU46XNRFSZ5QGHXY","created_at":"2026-08-11T02:24:38Z"},{"alias_kind":"pith_short_8","alias_value":"GU46XNRF","created_at":"2026-08-11T02:24:38Z"}],"graph_snapshots":[{"event_id":"sha256:92728fcaaa5d1a503501df73da7acb9417cdf90e5eb3ad21ae313c3762516147","target":"graph","created_at":"2026-08-11T02:24:38Z","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/2608.09624/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Internal safety scores judge a prompt before any text is generated, and they are validated by how well they separate harmful prompts from benign ones. That separation is then read as evidence that the score will also catch the attacks that succeed. Harmful intent is a property of the prompt. Jailbreak success is an outcome produced later by a particular target model, decoding policy, and judge. A filter tuned on a score that measures the wrong quantity spends its false positive budget on attacks that would have failed anyway. In this paper we audit that inference. Attention based measurements ","authors_text":"Ci Tao, Jun Dai, Ming Deng, Mingyu Luo, Ping Chen, Sijin Sun, Xiaoyan Sun, Xue Tan, Yangfu Li, Yiming Cheng, Zilang Qiu","cross_cats":["cs.AI","cs.CR"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-08-10T14:05:32Z","title":"Measuring the Wrong Thing: Internal Harmfulness Scores Anti-Rank Successful Jailbreaks"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.09624","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":"sha256:1e66439aafaea600acadda08bd462a71d757280490214081a0ace91a234ef580","target":"record","created_at":"2026-08-11T02:24:38Z","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":"788432c62332bf7dcdff48b971d36d1b43298771b265f4cb57878ab772149e78","cross_cats_sorted":["cs.AI","cs.CR"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-08-10T14:05:32Z","title_canon_sha256":"92b47b7bcfa958687ee20ca041649dda96012208251b9503d239fa1a02cb6945"},"schema_version":"1.0","source":{"id":"2608.09624","kind":"arxiv","version":1}},"canonical_sha256":"3539ebb625967b031ef834b4b1e4686e0d3f49fb35755ae837ae6f1dd2e94eb4","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3539ebb625967b031ef834b4b1e4686e0d3f49fb35755ae837ae6f1dd2e94eb4","first_computed_at":"2026-08-11T02:24:38.669806Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-08-11T02:24:38.669806Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"eewHBNHD8ZOFTcvQTJFpYD4AQyS30wLl6DPJKuH0zhGtzpB3Q52QHppCvcFprjAjXrI9PxWU9dqQ61I33Z9sCg==","signature_status":"signed_v1","signed_at":"2026-08-11T02:24:38.671488Z","signed_message":"canonical_sha256_bytes"},"source_id":"2608.09624","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1e66439aafaea600acadda08bd462a71d757280490214081a0ace91a234ef580","sha256:92728fcaaa5d1a503501df73da7acb9417cdf90e5eb3ad21ae313c3762516147"],"state_sha256":"648f68b111aa3772407477edeac364be13d939c00325cca2021579f03ee2ef56"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HNDPCQUYVXcveCiI/i6o0miGs5rbmCduaPxcBBmF4ux4cvnQtKGbNwi3ZgvaajBQ/DG7a5m3FuRg9TLG+nXdCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-12T22:19:55.793505Z","bundle_sha256":"8cfcf336d2b18c547ca96b7319b4c15541312653baee298ef0ee415181d7da24"}}