{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:GX4NI6W72P65FQJJIGXG6JQXNH","short_pith_number":"pith:GX4NI6W7","canonical_record":{"source":{"id":"2407.06866","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T13:53:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7bffb994498938f2e552ed6dcc5fca7e8349fd474520331e913579170ad40932","abstract_canon_sha256":"6d925abf61cce1ebd523703c3443dcf2960ed367edf17e463a6698a00e129ef9"},"schema_version":"1.0"},"canonical_sha256":"35f8d47adfd3fdd2c12941ae6f261769d5f339c2a81be57bd702df88703799b8","source":{"kind":"arxiv","id":"2407.06866","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.06866","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"arxiv_version","alias_value":"2407.06866v3","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.06866","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"pith_short_12","alias_value":"GX4NI6W72P65","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"pith_short_16","alias_value":"GX4NI6W72P65FQJJ","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"pith_short_8","alias_value":"GX4NI6W7","created_at":"2026-07-05T11:59:06Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:GX4NI6W72P65FQJJIGXG6JQXNH","target":"record","payload":{"canonical_record":{"source":{"id":"2407.06866","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T13:53:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7bffb994498938f2e552ed6dcc5fca7e8349fd474520331e913579170ad40932","abstract_canon_sha256":"6d925abf61cce1ebd523703c3443dcf2960ed367edf17e463a6698a00e129ef9"},"schema_version":"1.0"},"canonical_sha256":"35f8d47adfd3fdd2c12941ae6f261769d5f339c2a81be57bd702df88703799b8","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:59:06.090916Z","signature_b64":"4WYWuJw7buk24WQEs34fZjGl2yvrov/nmchWbaRnQpVJXjMJYZ0medt3kUptWLPyO6expOzEohOTeADtdn0LDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"35f8d47adfd3fdd2c12941ae6f261769d5f339c2a81be57bd702df88703799b8","last_reissued_at":"2026-07-05T11:59:06.090444Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:59:06.090444Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2407.06866","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:59:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Q3lJm2w9PkJyNhE9gpf+iIeQivk+BmblHlrcG7yc06yAi+YcMnTvpc4jbJbiCOXJ+ONA1J9UUm/lmYOllYb3Bg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-15T07:31:22.615296Z"},"content_sha256":"a4a84a9c92fdbaf071c193eb0086d4745c9162a158c97a1d78f565ae49e079f3","schema_version":"1.0","event_id":"sha256:a4a84a9c92fdbaf071c193eb0086d4745c9162a158c97a1d78f565ae49e079f3"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:GX4NI6W72P65FQJJIGXG6JQXNH","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"ChatGPT Doesn't Trust Chargers Fans: Guardrail Sensitivity in Context","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Naomi Saphra, Victoria R. Li, Yida Chen","submitted_at":"2024-07-09T13:53:38Z","abstract_excerpt":"While the biases of language models in production are extensively documented, the biases of their guardrails have been neglected. This paper studies how contextual information about the user influences the likelihood of an LLM to refuse to execute a request. By generating user biographies that offer ideological and demographic information, we find a number of biases in guardrail sensitivity on GPT-3.5. Younger, female, and Asian-American personas are more likely to trigger a refusal guardrail when requesting censored or illegal information. Guardrails are also sycophantic, refusing to comply w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.06866","kind":"arxiv","version":3},"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/2407.06866/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":"2026-07-05T11:59:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lbeXg57cyO+qQk/VMSMhKfIYSuSO0VWkh+PF/h+TibgFu8Qrnau/5WOvxMbXW9XToHrmudo1Yu8Cpdhpd/h+BA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-15T07:31:22.616141Z"},"content_sha256":"aba83f49fe37ec0479646c86bd421c866cbbe5a3600a5a1016693d789a3e3437","schema_version":"1.0","event_id":"sha256:aba83f49fe37ec0479646c86bd421c866cbbe5a3600a5a1016693d789a3e3437"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/GX4NI6W72P65FQJJIGXG6JQXNH/bundle.json","state_url":"https://pith.science/pith/GX4NI6W72P65FQJJIGXG6JQXNH/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/GX4NI6W72P65FQJJIGXG6JQXNH/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":"2026-08-15T07:31:22Z","links":{"resolver":"https://pith.science/pith/GX4NI6W72P65FQJJIGXG6JQXNH","bundle":"https://pith.science/pith/GX4NI6W72P65FQJJIGXG6JQXNH/bundle.json","state":"https://pith.science/pith/GX4NI6W72P65FQJJIGXG6JQXNH/state.json","well_known_bundle":"https://pith.science/.well-known/pith/GX4NI6W72P65FQJJIGXG6JQXNH/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:GX4NI6W72P65FQJJIGXG6JQXNH","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":"6d925abf61cce1ebd523703c3443dcf2960ed367edf17e463a6698a00e129ef9","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T13:53:38Z","title_canon_sha256":"7bffb994498938f2e552ed6dcc5fca7e8349fd474520331e913579170ad40932"},"schema_version":"1.0","source":{"id":"2407.06866","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.06866","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"arxiv_version","alias_value":"2407.06866v3","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.06866","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"pith_short_12","alias_value":"GX4NI6W72P65","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"pith_short_16","alias_value":"GX4NI6W72P65FQJJ","created_at":"2026-07-05T11:59:06Z"},{"alias_kind":"pith_short_8","alias_value":"GX4NI6W7","created_at":"2026-07-05T11:59:06Z"}],"graph_snapshots":[{"event_id":"sha256:aba83f49fe37ec0479646c86bd421c866cbbe5a3600a5a1016693d789a3e3437","target":"graph","created_at":"2026-07-05T11:59:06Z","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/2407.06866/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"While the biases of language models in production are extensively documented, the biases of their guardrails have been neglected. This paper studies how contextual information about the user influences the likelihood of an LLM to refuse to execute a request. By generating user biographies that offer ideological and demographic information, we find a number of biases in guardrail sensitivity on GPT-3.5. Younger, female, and Asian-American personas are more likely to trigger a refusal guardrail when requesting censored or illegal information. Guardrails are also sycophantic, refusing to comply w","authors_text":"Naomi Saphra, Victoria R. Li, Yida Chen","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T13:53:38Z","title":"ChatGPT Doesn't Trust Chargers Fans: Guardrail Sensitivity in Context"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.06866","kind":"arxiv","version":3},"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:a4a84a9c92fdbaf071c193eb0086d4745c9162a158c97a1d78f565ae49e079f3","target":"record","created_at":"2026-07-05T11:59:06Z","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":"6d925abf61cce1ebd523703c3443dcf2960ed367edf17e463a6698a00e129ef9","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T13:53:38Z","title_canon_sha256":"7bffb994498938f2e552ed6dcc5fca7e8349fd474520331e913579170ad40932"},"schema_version":"1.0","source":{"id":"2407.06866","kind":"arxiv","version":3}},"canonical_sha256":"35f8d47adfd3fdd2c12941ae6f261769d5f339c2a81be57bd702df88703799b8","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"35f8d47adfd3fdd2c12941ae6f261769d5f339c2a81be57bd702df88703799b8","first_computed_at":"2026-07-05T11:59:06.090444Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:59:06.090444Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"4WYWuJw7buk24WQEs34fZjGl2yvrov/nmchWbaRnQpVJXjMJYZ0medt3kUptWLPyO6expOzEohOTeADtdn0LDA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:59:06.090916Z","signed_message":"canonical_sha256_bytes"},"source_id":"2407.06866","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a4a84a9c92fdbaf071c193eb0086d4745c9162a158c97a1d78f565ae49e079f3","sha256:aba83f49fe37ec0479646c86bd421c866cbbe5a3600a5a1016693d789a3e3437"],"state_sha256":"be911f27f21ab70ac5e8a10f60f281759174a442a8b36013ecc215ac75196950"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QdY0PzZnfi0V1O4aDOoswNWGQ0EsiJ5fpNhjzGikF4HLRw4YL6QLBGGO0f0GVJWurE4tDmJMA4j9eUa2T2hBAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-15T07:31:22.624314Z","bundle_sha256":"e10ec3102e11565bc3aa1a4187ade85e69b0fbad3098ff39ec2d76fd9bcc606e"}}