{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:X4CQHGYZV6BYQFURDAP752SGT4","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":"02ceab85d42ad8cb4ca92d22283738dc476be4751a881cfb37ab07b95f83197e","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-08-22T09:57:57Z","title_canon_sha256":"2da71973c28d344ac95e62d3d532bf2060d5fbe701c70305f8c6f92d8207c0f5"},"schema_version":"1.0","source":{"id":"2408.12259","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2408.12259","created_at":"2026-07-05T10:13:26Z"},{"alias_kind":"arxiv_version","alias_value":"2408.12259v2","created_at":"2026-07-05T10:13:26Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.12259","created_at":"2026-07-05T10:13:26Z"},{"alias_kind":"pith_short_12","alias_value":"X4CQHGYZV6BY","created_at":"2026-07-05T10:13:26Z"},{"alias_kind":"pith_short_16","alias_value":"X4CQHGYZV6BYQFUR","created_at":"2026-07-05T10:13:26Z"},{"alias_kind":"pith_short_8","alias_value":"X4CQHGYZ","created_at":"2026-07-05T10:13:26Z"}],"graph_snapshots":[{"event_id":"sha256:dd6ab3eb2dba5574534d5d08a8d05d5662cfe8407e49d7a0ec4fe378dd54242a","target":"graph","created_at":"2026-07-05T10:13:26Z","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/2408.12259/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Consider a scenario where a harmfulness evaluation metric intended to filter unsafe responses from a Large Language Model. When applied to individual harmful prompt-response pairs, it correctly flags them as unsafe by assigning a high-risk score. Yet, if those same pairs are concatenated, the metrics decision unexpectedly reverses - labelling the combined content as safe with a low score, allowing the harmful text to bypass the filter. We found that multiple safety metrics, including advanced metrics such as GPT-based judges, exhibit this non-safe behaviour. Moreover, they show a strong sensit","authors_text":"Eitan Farchi, George Kour, Leshem Choshen, Ora Nova Fandina, Orna Raz, Yotam Perlitz","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-08-22T09:57:57Z","title":"How Safe is Your Safety Metric? Automatic Concatenation Tests for Metric Reliability"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.12259","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":"sha256:2bdb0302a0893e4a84854ccab2b86b057f5c711f54345528b976a7de11dcf4fd","target":"record","created_at":"2026-07-05T10:13:26Z","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":"02ceab85d42ad8cb4ca92d22283738dc476be4751a881cfb37ab07b95f83197e","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-08-22T09:57:57Z","title_canon_sha256":"2da71973c28d344ac95e62d3d532bf2060d5fbe701c70305f8c6f92d8207c0f5"},"schema_version":"1.0","source":{"id":"2408.12259","kind":"arxiv","version":2}},"canonical_sha256":"bf05039b19af83881691181ffeea469f1fbb8ecb9e2ab37e1ffdb9df03fb76a5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bf05039b19af83881691181ffeea469f1fbb8ecb9e2ab37e1ffdb9df03fb76a5","first_computed_at":"2026-07-05T10:13:26.886024Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:13:26.886024Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"vShT3crQ7Jmhf2pzSe5I3Q2o78tyuDiA/vq8OAUD9WXDibyxoY2vSQOforeYlXlx3FteqBCgO0bKetdbsbQ6AA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:13:26.886643Z","signed_message":"canonical_sha256_bytes"},"source_id":"2408.12259","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:2bdb0302a0893e4a84854ccab2b86b057f5c711f54345528b976a7de11dcf4fd","sha256:dd6ab3eb2dba5574534d5d08a8d05d5662cfe8407e49d7a0ec4fe378dd54242a"],"state_sha256":"945073f55f8a887f241353422288dcb5eb37c47d16edaf8b0ff5b267ff29ee34"}