{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:PQOHUFUIDPRTGAU36WHAAGVPMN","short_pith_number":"pith:PQOHUFUI","canonical_record":{"source":{"id":"2506.18032","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-06-22T13:27:09Z","cross_cats_sorted":[],"title_canon_sha256":"f1c50e93b0d63412a538156227e8b931c696ba4a23ca6b3ea561e5eb0f6aac30","abstract_canon_sha256":"25ffcbda04641edec043b8c1bb0df1d974c059dd08848a118fed5618dde5dfb0"},"schema_version":"1.0"},"canonical_sha256":"7c1c7a16881be333029bf58e001aaf63401dd87a866b0d0bf12d735c9a0ee77e","source":{"kind":"arxiv","id":"2506.18032","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.18032","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"arxiv_version","alias_value":"2506.18032v1","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.18032","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"pith_short_12","alias_value":"PQOHUFUIDPRT","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"pith_short_16","alias_value":"PQOHUFUIDPRTGAU3","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"pith_short_8","alias_value":"PQOHUFUI","created_at":"2026-07-05T11:25:27Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:PQOHUFUIDPRTGAU36WHAAGVPMN","target":"record","payload":{"canonical_record":{"source":{"id":"2506.18032","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-06-22T13:27:09Z","cross_cats_sorted":[],"title_canon_sha256":"f1c50e93b0d63412a538156227e8b931c696ba4a23ca6b3ea561e5eb0f6aac30","abstract_canon_sha256":"25ffcbda04641edec043b8c1bb0df1d974c059dd08848a118fed5618dde5dfb0"},"schema_version":"1.0"},"canonical_sha256":"7c1c7a16881be333029bf58e001aaf63401dd87a866b0d0bf12d735c9a0ee77e","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:25:27.374812Z","signature_b64":"wVbTz+flzn234hv01D1goFKLX59V/DpyBAjweaW2J9N9g/djmXeljTfEkfYhQERmm1UXRC340HqXGxPh9sBuAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7c1c7a16881be333029bf58e001aaf63401dd87a866b0d0bf12d735c9a0ee77e","last_reissued_at":"2026-07-05T11:25:27.374297Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:25:27.374297Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2506.18032","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":"2026-07-05T11:25:27Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"XyMqpF7DIfYvRQvhCYkrGYK2KBFvojTPDVCGrKjgpa42GjkcfyeCxdFD4mMYTP9ppBpYXcGeP9Nspb4S52wlDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T18:47:10.057884Z"},"content_sha256":"9ceef9e5fd5fb376f8192bba2ad1ba10a70294fb38b923c3d98e1e6513882473","schema_version":"1.0","event_id":"sha256:9ceef9e5fd5fb376f8192bba2ad1ba10a70294fb38b923c3d98e1e6513882473"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:PQOHUFUIDPRTGAU36WHAAGVPMN","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Why Do Some Language Models Fake Alignment While Others Don't?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Abhay Sheshadri, Alex Mallen, Arun Jose, Fabien Roger, Janus, John Hughes, Julian Michael","submitted_at":"2025-06-22T13:27:09Z","abstract_excerpt":"Alignment faking in large language models presented a demonstration of Claude 3 Opus and Claude 3.5 Sonnet selectively complying with a helpful-only training objective to prevent modification of their behavior outside of training. We expand this analysis to 25 models and find that only 5 (Claude 3 Opus, Claude 3.5 Sonnet, Llama 3 405B, Grok 3, Gemini 2.0 Flash) comply with harmful queries more when they infer they are in training than when they infer they are in deployment. First, we study the motivations of these 5 models. Results from perturbing details of the scenario suggest that only Clau"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.18032","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/2506.18032/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":"2026-07-05T11:25:27Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1JFql05kWsXt3VSMg8KpuQ1hETu2ARpE4SUDMP0Kd9V7/GU0OyM3zhmF17yZSSYGgWNe9mGsA65tmxicSH9VCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T18:47:10.058406Z"},"content_sha256":"0ec094999463ca73f0296bef810b435bb6e68b01b93a82fe8e5705ec0ad9423d","schema_version":"1.0","event_id":"sha256:0ec094999463ca73f0296bef810b435bb6e68b01b93a82fe8e5705ec0ad9423d"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/PQOHUFUIDPRTGAU36WHAAGVPMN/bundle.json","state_url":"https://pith.science/pith/PQOHUFUIDPRTGAU36WHAAGVPMN/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/PQOHUFUIDPRTGAU36WHAAGVPMN/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":"2026-08-07T18:47:10Z","links":{"resolver":"https://pith.science/pith/PQOHUFUIDPRTGAU36WHAAGVPMN","bundle":"https://pith.science/pith/PQOHUFUIDPRTGAU36WHAAGVPMN/bundle.json","state":"https://pith.science/pith/PQOHUFUIDPRTGAU36WHAAGVPMN/state.json","well_known_bundle":"https://pith.science/.well-known/pith/PQOHUFUIDPRTGAU36WHAAGVPMN/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:PQOHUFUIDPRTGAU36WHAAGVPMN","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":"25ffcbda04641edec043b8c1bb0df1d974c059dd08848a118fed5618dde5dfb0","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-06-22T13:27:09Z","title_canon_sha256":"f1c50e93b0d63412a538156227e8b931c696ba4a23ca6b3ea561e5eb0f6aac30"},"schema_version":"1.0","source":{"id":"2506.18032","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.18032","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"arxiv_version","alias_value":"2506.18032v1","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.18032","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"pith_short_12","alias_value":"PQOHUFUIDPRT","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"pith_short_16","alias_value":"PQOHUFUIDPRTGAU3","created_at":"2026-07-05T11:25:27Z"},{"alias_kind":"pith_short_8","alias_value":"PQOHUFUI","created_at":"2026-07-05T11:25:27Z"}],"graph_snapshots":[{"event_id":"sha256:0ec094999463ca73f0296bef810b435bb6e68b01b93a82fe8e5705ec0ad9423d","target":"graph","created_at":"2026-07-05T11:25:27Z","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/2506.18032/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Alignment faking in large language models presented a demonstration of Claude 3 Opus and Claude 3.5 Sonnet selectively complying with a helpful-only training objective to prevent modification of their behavior outside of training. We expand this analysis to 25 models and find that only 5 (Claude 3 Opus, Claude 3.5 Sonnet, Llama 3 405B, Grok 3, Gemini 2.0 Flash) comply with harmful queries more when they infer they are in training than when they infer they are in deployment. First, we study the motivations of these 5 models. Results from perturbing details of the scenario suggest that only Clau","authors_text":"Abhay Sheshadri, Alex Mallen, Arun Jose, Fabien Roger, Janus, John Hughes, Julian Michael","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-06-22T13:27:09Z","title":"Why Do Some Language Models Fake Alignment While Others Don't?"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.18032","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":"sha256:9ceef9e5fd5fb376f8192bba2ad1ba10a70294fb38b923c3d98e1e6513882473","target":"record","created_at":"2026-07-05T11:25:27Z","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":"25ffcbda04641edec043b8c1bb0df1d974c059dd08848a118fed5618dde5dfb0","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-06-22T13:27:09Z","title_canon_sha256":"f1c50e93b0d63412a538156227e8b931c696ba4a23ca6b3ea561e5eb0f6aac30"},"schema_version":"1.0","source":{"id":"2506.18032","kind":"arxiv","version":1}},"canonical_sha256":"7c1c7a16881be333029bf58e001aaf63401dd87a866b0d0bf12d735c9a0ee77e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"7c1c7a16881be333029bf58e001aaf63401dd87a866b0d0bf12d735c9a0ee77e","first_computed_at":"2026-07-05T11:25:27.374297Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:25:27.374297Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"wVbTz+flzn234hv01D1goFKLX59V/DpyBAjweaW2J9N9g/djmXeljTfEkfYhQERmm1UXRC340HqXGxPh9sBuAg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:25:27.374812Z","signed_message":"canonical_sha256_bytes"},"source_id":"2506.18032","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:9ceef9e5fd5fb376f8192bba2ad1ba10a70294fb38b923c3d98e1e6513882473","sha256:0ec094999463ca73f0296bef810b435bb6e68b01b93a82fe8e5705ec0ad9423d"],"state_sha256":"d1dd765c54d1cd6625e3755217480ec02cbddd5b1599c800c30eae68e344f4dd"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lQcowPlIJqn98BwiIrP2JpulWwPDoT6iEtiI6Gry+dWd679nFwVmTyGK0k7ac7Hehx6FU9jBhEy0I94Fjw2OCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T18:47:10.062709Z","bundle_sha256":"e109b9712e850b2c3ad2d6bd1fecf11789546aa5244967a3fd4ed8aa7f9f98ae"}}