{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:AIIP4JUF5URVIIGQWRAQ2HN4CB","short_pith_number":"pith:AIIP4JUF","canonical_record":{"source":{"id":"2405.05466","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-05-08T23:44:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e8d2c463987412d87bf7ca17e0da69a7a71b7b3fdebc76ac95403de477ec0c18","abstract_canon_sha256":"c581c454a80d2b37582b5820ecfe0d8aea65bbf4b2290b46ed142a76b577ff30"},"schema_version":"1.0"},"canonical_sha256":"0210fe2685ed235420d0b4410d1dbc10644468d71198367b91263db21788ecf2","source":{"kind":"arxiv","id":"2405.05466","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.05466","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"arxiv_version","alias_value":"2405.05466v2","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.05466","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"pith_short_12","alias_value":"AIIP4JUF5URV","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"pith_short_16","alias_value":"AIIP4JUF5URVIIGQ","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"pith_short_8","alias_value":"AIIP4JUF","created_at":"2026-07-05T08:18:09Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:AIIP4JUF5URVIIGQWRAQ2HN4CB","target":"record","payload":{"canonical_record":{"source":{"id":"2405.05466","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-05-08T23:44:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e8d2c463987412d87bf7ca17e0da69a7a71b7b3fdebc76ac95403de477ec0c18","abstract_canon_sha256":"c581c454a80d2b37582b5820ecfe0d8aea65bbf4b2290b46ed142a76b577ff30"},"schema_version":"1.0"},"canonical_sha256":"0210fe2685ed235420d0b4410d1dbc10644468d71198367b91263db21788ecf2","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:18:09.045878Z","signature_b64":"3LYNm0h/SU9vZJ5Gzl/woArhRRt2E1aXz1YP3Jx3PkhPJt5+Mpt7ZskJcR8kQYKz0+Qyx8kJ3rF8OfcGXutJBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0210fe2685ed235420d0b4410d1dbc10644468d71198367b91263db21788ecf2","last_reissued_at":"2026-07-05T08:18:09.045433Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:18:09.045433Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2405.05466","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":"2026-07-05T08:18:09Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"kOS6pO8FIfxIxrAK2eJhjt+2m+MJ3N73fra4ljIHGIKkyVjRyasPwBygFSU3USnzXFjvHp1xRhNxgLmemsooAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T06:25:47.210862Z"},"content_sha256":"ae2788ce000d5ff567b136a31250e68b1ba6ca582b078825c38c6ef0088d89c1","schema_version":"1.0","event_id":"sha256:ae2788ce000d5ff567b136a31250e68b1ba6ca582b078825c38c6ef0088d89c1"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:AIIP4JUF5URVIIGQWRAQ2HN4CB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Poser: Unmasking Alignment Faking LLMs by Manipulating Their Internals","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Caden Juang, Joshua Clymer, Severin Field","submitted_at":"2024-05-08T23:44:08Z","abstract_excerpt":"Like a criminal under investigation, Large Language Models (LLMs) might pretend to be aligned while evaluated and misbehave when they have a good opportunity. Can current interpretability methods catch these 'alignment fakers?' To answer this question, we introduce a benchmark that consists of 324 pairs of LLMs fine-tuned to select actions in role-play scenarios. One model in each pair is consistently benign (aligned). The other model misbehaves in scenarios where it is unlikely to be caught (alignment faking). The task is to identify the alignment faking model using only inputs where the two "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.05466","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/2405.05466/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":"2026-07-05T08:18:09Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wwEPmdpVdZhsentkNtRJq5Amy8hi7PwYAuA8DghBC7BXovsT3yeSbJUF9Rc+HJ7Y76phGF65MugYHF7M6EILCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T06:25:47.214654Z"},"content_sha256":"5c913c188a86df4a57cbc524a7aa385a4a57f529c1b9447d82c96ab03416c81c","schema_version":"1.0","event_id":"sha256:5c913c188a86df4a57cbc524a7aa385a4a57f529c1b9447d82c96ab03416c81c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/AIIP4JUF5URVIIGQWRAQ2HN4CB/bundle.json","state_url":"https://pith.science/pith/AIIP4JUF5URVIIGQWRAQ2HN4CB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/AIIP4JUF5URVIIGQWRAQ2HN4CB/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":"2026-08-06T06:25:47Z","links":{"resolver":"https://pith.science/pith/AIIP4JUF5URVIIGQWRAQ2HN4CB","bundle":"https://pith.science/pith/AIIP4JUF5URVIIGQWRAQ2HN4CB/bundle.json","state":"https://pith.science/pith/AIIP4JUF5URVIIGQWRAQ2HN4CB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/AIIP4JUF5URVIIGQWRAQ2HN4CB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:AIIP4JUF5URVIIGQWRAQ2HN4CB","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":"c581c454a80d2b37582b5820ecfe0d8aea65bbf4b2290b46ed142a76b577ff30","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-05-08T23:44:08Z","title_canon_sha256":"e8d2c463987412d87bf7ca17e0da69a7a71b7b3fdebc76ac95403de477ec0c18"},"schema_version":"1.0","source":{"id":"2405.05466","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.05466","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"arxiv_version","alias_value":"2405.05466v2","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.05466","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"pith_short_12","alias_value":"AIIP4JUF5URV","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"pith_short_16","alias_value":"AIIP4JUF5URVIIGQ","created_at":"2026-07-05T08:18:09Z"},{"alias_kind":"pith_short_8","alias_value":"AIIP4JUF","created_at":"2026-07-05T08:18:09Z"}],"graph_snapshots":[{"event_id":"sha256:5c913c188a86df4a57cbc524a7aa385a4a57f529c1b9447d82c96ab03416c81c","target":"graph","created_at":"2026-07-05T08:18:09Z","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/2405.05466/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Like a criminal under investigation, Large Language Models (LLMs) might pretend to be aligned while evaluated and misbehave when they have a good opportunity. Can current interpretability methods catch these 'alignment fakers?' To answer this question, we introduce a benchmark that consists of 324 pairs of LLMs fine-tuned to select actions in role-play scenarios. One model in each pair is consistently benign (aligned). The other model misbehaves in scenarios where it is unlikely to be caught (alignment faking). The task is to identify the alignment faking model using only inputs where the two ","authors_text":"Caden Juang, Joshua Clymer, Severin Field","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-05-08T23:44:08Z","title":"Poser: Unmasking Alignment Faking LLMs by Manipulating Their Internals"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.05466","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:ae2788ce000d5ff567b136a31250e68b1ba6ca582b078825c38c6ef0088d89c1","target":"record","created_at":"2026-07-05T08:18:09Z","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":"c581c454a80d2b37582b5820ecfe0d8aea65bbf4b2290b46ed142a76b577ff30","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-05-08T23:44:08Z","title_canon_sha256":"e8d2c463987412d87bf7ca17e0da69a7a71b7b3fdebc76ac95403de477ec0c18"},"schema_version":"1.0","source":{"id":"2405.05466","kind":"arxiv","version":2}},"canonical_sha256":"0210fe2685ed235420d0b4410d1dbc10644468d71198367b91263db21788ecf2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"0210fe2685ed235420d0b4410d1dbc10644468d71198367b91263db21788ecf2","first_computed_at":"2026-07-05T08:18:09.045433Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:18:09.045433Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"3LYNm0h/SU9vZJ5Gzl/woArhRRt2E1aXz1YP3Jx3PkhPJt5+Mpt7ZskJcR8kQYKz0+Qyx8kJ3rF8OfcGXutJBg==","signature_status":"signed_v1","signed_at":"2026-07-05T08:18:09.045878Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.05466","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ae2788ce000d5ff567b136a31250e68b1ba6ca582b078825c38c6ef0088d89c1","sha256:5c913c188a86df4a57cbc524a7aa385a4a57f529c1b9447d82c96ab03416c81c"],"state_sha256":"e561b849b7133c6ab3a9fdc4d636571fe4eae33e8d5ca9a9832798e4a6cefd1e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7ltx9EuSkqrVGB87QFmp2+kYoU3hULT67xEC29XPe/yhxfxWqG3k2Iw3QEkfbC3ptY+xSy4G1xGW7JWcG72/BQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T06:25:47.221920Z","bundle_sha256":"98b80fef1dd28388be7ad6c39a6b1a96005a2efc36dd22673d5b090a90f7067a"}}