{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:AAHYCYP3MKTDEGISAFEGYN3RHQ","short_pith_number":"pith:AAHYCYP3","canonical_record":{"source":{"id":"2502.16797","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T03:16:15Z","cross_cats_sorted":[],"title_canon_sha256":"a820946f70b2fa6f9a92b9bf98c36081d9af768908ee6b90f6270c151d275fad","abstract_canon_sha256":"6f3f6a0a878587f1c17befe56c6eb5954557778e6605c969a914245e1cf6a861"},"schema_version":"1.0"},"canonical_sha256":"000f8161fb62a632191201486c37713c2b3df0d6a48fc7bdfd9c226776c37315","source":{"kind":"arxiv","id":"2502.16797","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.16797","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"arxiv_version","alias_value":"2502.16797v1","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.16797","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"pith_short_12","alias_value":"AAHYCYP3MKTD","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"pith_short_16","alias_value":"AAHYCYP3MKTDEGIS","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"pith_short_8","alias_value":"AAHYCYP3","created_at":"2026-07-05T10:18:59Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:AAHYCYP3MKTDEGISAFEGYN3RHQ","target":"record","payload":{"canonical_record":{"source":{"id":"2502.16797","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T03:16:15Z","cross_cats_sorted":[],"title_canon_sha256":"a820946f70b2fa6f9a92b9bf98c36081d9af768908ee6b90f6270c151d275fad","abstract_canon_sha256":"6f3f6a0a878587f1c17befe56c6eb5954557778e6605c969a914245e1cf6a861"},"schema_version":"1.0"},"canonical_sha256":"000f8161fb62a632191201486c37713c2b3df0d6a48fc7bdfd9c226776c37315","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:18:59.799614Z","signature_b64":"6KQ8N92uu0nrgoVqAMJFgauKHxzY7LbptQvtrsCqKtLnMIpr5Tejg9guc19MKgFZwmCUfZNcWVoc+ydcd/CfBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"000f8161fb62a632191201486c37713c2b3df0d6a48fc7bdfd9c226776c37315","last_reissued_at":"2026-07-05T10:18:59.799123Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:18:59.799123Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2502.16797","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":"2026-07-05T10:18:59Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rdJw0xj8Ro3mfRpDMYYSSDj3Hy/4lzE480qPIKTfpK2CgnZzhmPVnUH8qPr0XJ9t0LbUP3MMg4v1Sr7HmxkSAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T11:24:37.684144Z"},"content_sha256":"834be4a9c642edff5ec8a4dc9c6aee74a120c5628576d8a9f762186d1149fea8","schema_version":"1.0","event_id":"sha256:834be4a9c642edff5ec8a4dc9c6aee74a120c5628576d8a9f762186d1149fea8"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:AAHYCYP3MKTDEGISAFEGYN3RHQ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Forecasting Rare Language Model Behaviors","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Erik Jones, Ethan Perez, Jan Leike, Jared Kaplan, Jesse Mu, Meg Tong, Mohammed Mahfoud, Mrinank Sharma, Roger Grosse, William Fithian","submitted_at":"2025-02-24T03:16:15Z","abstract_excerpt":"Standard language model evaluations can fail to capture risks that emerge only at deployment scale. For example, a model may produce safe responses during a small-scale beta test, yet reveal dangerous information when processing billions of requests at deployment. To remedy this, we introduce a method to forecast potential risks across orders of magnitude more queries than we test during evaluation. We make forecasts by studying each query's elicitation probability -- the probability the query produces a target behavior -- and demonstrate that the largest observed elicitation probabilities pre"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.16797","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/2502.16797/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":"2026-07-05T10:18:59Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"302tY0NvAiXnXGx/Z5bSOYdGSmUBegxpAr7K/M9r+B7LpqMBGc8G+j6zNx+WBiHTDetjnCGfVFVO18gI+Vm0Dg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T11:24:37.684835Z"},"content_sha256":"8fcfb20bff217d5e1abff7e124dd5ecc382ec49d41a386ca1b35ad390e5b5609","schema_version":"1.0","event_id":"sha256:8fcfb20bff217d5e1abff7e124dd5ecc382ec49d41a386ca1b35ad390e5b5609"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/AAHYCYP3MKTDEGISAFEGYN3RHQ/bundle.json","state_url":"https://pith.science/pith/AAHYCYP3MKTDEGISAFEGYN3RHQ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/AAHYCYP3MKTDEGISAFEGYN3RHQ/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":"2026-08-04T11:24:37Z","links":{"resolver":"https://pith.science/pith/AAHYCYP3MKTDEGISAFEGYN3RHQ","bundle":"https://pith.science/pith/AAHYCYP3MKTDEGISAFEGYN3RHQ/bundle.json","state":"https://pith.science/pith/AAHYCYP3MKTDEGISAFEGYN3RHQ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/AAHYCYP3MKTDEGISAFEGYN3RHQ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:AAHYCYP3MKTDEGISAFEGYN3RHQ","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":"6f3f6a0a878587f1c17befe56c6eb5954557778e6605c969a914245e1cf6a861","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T03:16:15Z","title_canon_sha256":"a820946f70b2fa6f9a92b9bf98c36081d9af768908ee6b90f6270c151d275fad"},"schema_version":"1.0","source":{"id":"2502.16797","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.16797","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"arxiv_version","alias_value":"2502.16797v1","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.16797","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"pith_short_12","alias_value":"AAHYCYP3MKTD","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"pith_short_16","alias_value":"AAHYCYP3MKTDEGIS","created_at":"2026-07-05T10:18:59Z"},{"alias_kind":"pith_short_8","alias_value":"AAHYCYP3","created_at":"2026-07-05T10:18:59Z"}],"graph_snapshots":[{"event_id":"sha256:8fcfb20bff217d5e1abff7e124dd5ecc382ec49d41a386ca1b35ad390e5b5609","target":"graph","created_at":"2026-07-05T10:18:59Z","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/2502.16797/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Standard language model evaluations can fail to capture risks that emerge only at deployment scale. For example, a model may produce safe responses during a small-scale beta test, yet reveal dangerous information when processing billions of requests at deployment. To remedy this, we introduce a method to forecast potential risks across orders of magnitude more queries than we test during evaluation. We make forecasts by studying each query's elicitation probability -- the probability the query produces a target behavior -- and demonstrate that the largest observed elicitation probabilities pre","authors_text":"Erik Jones, Ethan Perez, Jan Leike, Jared Kaplan, Jesse Mu, Meg Tong, Mohammed Mahfoud, Mrinank Sharma, Roger Grosse, William Fithian","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T03:16:15Z","title":"Forecasting Rare Language Model Behaviors"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.16797","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":"sha256:834be4a9c642edff5ec8a4dc9c6aee74a120c5628576d8a9f762186d1149fea8","target":"record","created_at":"2026-07-05T10:18:59Z","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":"6f3f6a0a878587f1c17befe56c6eb5954557778e6605c969a914245e1cf6a861","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T03:16:15Z","title_canon_sha256":"a820946f70b2fa6f9a92b9bf98c36081d9af768908ee6b90f6270c151d275fad"},"schema_version":"1.0","source":{"id":"2502.16797","kind":"arxiv","version":1}},"canonical_sha256":"000f8161fb62a632191201486c37713c2b3df0d6a48fc7bdfd9c226776c37315","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"000f8161fb62a632191201486c37713c2b3df0d6a48fc7bdfd9c226776c37315","first_computed_at":"2026-07-05T10:18:59.799123Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:18:59.799123Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"6KQ8N92uu0nrgoVqAMJFgauKHxzY7LbptQvtrsCqKtLnMIpr5Tejg9guc19MKgFZwmCUfZNcWVoc+ydcd/CfBA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:18:59.799614Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.16797","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:834be4a9c642edff5ec8a4dc9c6aee74a120c5628576d8a9f762186d1149fea8","sha256:8fcfb20bff217d5e1abff7e124dd5ecc382ec49d41a386ca1b35ad390e5b5609"],"state_sha256":"d1474811006a6e4895e31f1ebe303569800aa464a7e7f1ac6f1e763c749e2587"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"2hRjdEKaAv5zHpkZjv1t+LflOIxsUmDlkAeNNIwg60CLtO6IpbS8FEo2i5m8BbiGS1M6mQm5i36sCuwWzRXYAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-04T11:24:37.690295Z","bundle_sha256":"4105224962e591324531b09a7307340f6c3abb3f048a9f9b0464251b6475cec8"}}