{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:J6VHBVZQ7JCASNI6CG352I6WF6","short_pith_number":"pith:J6VHBVZQ","canonical_record":{"source":{"id":"2504.06577","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-09T04:58:14Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"43be7874230e6864badccc0bfbe717a268a63fbb5d1669775c8edad857e18cc6","abstract_canon_sha256":"aa00f64db9a6b6b71773a2aea7fd827867e5ec3fd65b7e65532ccd483aff4717"},"schema_version":"1.0"},"canonical_sha256":"4faa70d730fa4409351e11b7dd23d62f8a6d86516039791d9dae5ff3d75d3092","source":{"kind":"arxiv","id":"2504.06577","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.06577","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"arxiv_version","alias_value":"2504.06577v1","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.06577","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"pith_short_12","alias_value":"J6VHBVZQ7JCA","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"pith_short_16","alias_value":"J6VHBVZQ7JCASNI6","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"pith_short_8","alias_value":"J6VHBVZQ","created_at":"2026-07-05T10:46:29Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:J6VHBVZQ7JCASNI6CG352I6WF6","target":"record","payload":{"canonical_record":{"source":{"id":"2504.06577","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-09T04:58:14Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"43be7874230e6864badccc0bfbe717a268a63fbb5d1669775c8edad857e18cc6","abstract_canon_sha256":"aa00f64db9a6b6b71773a2aea7fd827867e5ec3fd65b7e65532ccd483aff4717"},"schema_version":"1.0"},"canonical_sha256":"4faa70d730fa4409351e11b7dd23d62f8a6d86516039791d9dae5ff3d75d3092","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:46:29.190062Z","signature_b64":"bNz+qCmWe8lrPL8MKiOG8iYU5/cazIwO9ZLy+Il7R1e+8f1T8ILb1qQsX+gLXjD8pOpKDS6OfCLjJCKzhZiKAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4faa70d730fa4409351e11b7dd23d62f8a6d86516039791d9dae5ff3d75d3092","last_reissued_at":"2026-07-05T10:46:29.189624Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:46:29.189624Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2504.06577","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":"2026-07-05T10:46:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6ntvYgj+KC37zlWWjkNP/uctDQBRIc3wM4JhmqRDwC4kLuOEPS2lwzyCHDvj9uCNGSnXtM7pHA22C9zcm/n/AA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-18T19:51:41.785696Z"},"content_sha256":"6bb6b1d3bc92a24d07c4358a49c8249d86f052ab808c4a07a2bf58300fe75853","schema_version":"1.0","event_id":"sha256:6bb6b1d3bc92a24d07c4358a49c8249d86f052ab808c4a07a2bf58300fe75853"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:J6VHBVZQ7JCASNI6CG352I6WF6","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Bypassing Safety Guardrails in LLMs Using Humor","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Pedro Cisneros-Velarde","submitted_at":"2025-04-09T04:58:14Z","abstract_excerpt":"In this paper, we show it is possible to bypass the safety guardrails of large language models (LLMs) through a humorous prompt including the unsafe request. In particular, our method does not edit the unsafe request and follows a fixed template -- it is simple to implement and does not need additional LLMs to craft prompts. Extensive experiments show the effectiveness of our method across different LLMs. We also show that both removing and adding more humor to our method can reduce its effectiveness -- excessive humor possibly distracts the LLM from fulfilling its unsafe request. Thus, we arg"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.06577","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/2504.06577/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":"2026-07-05T10:46:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UFbCpcVGgH8r57QiNTiSl5P8eEwBRiz+OldJkCuxxH3avePYk54xwjsaj9Z87gAN2urhVdNjjGyOK8ILMdf8CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-18T19:51:41.786235Z"},"content_sha256":"192c8306f8d364cc22fd554d183a13925f3ff4fb16b93bd91df921d9bcc7adac","schema_version":"1.0","event_id":"sha256:192c8306f8d364cc22fd554d183a13925f3ff4fb16b93bd91df921d9bcc7adac"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/J6VHBVZQ7JCASNI6CG352I6WF6/bundle.json","state_url":"https://pith.science/pith/J6VHBVZQ7JCASNI6CG352I6WF6/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/J6VHBVZQ7JCASNI6CG352I6WF6/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":"2026-08-18T19:51:41Z","links":{"resolver":"https://pith.science/pith/J6VHBVZQ7JCASNI6CG352I6WF6","bundle":"https://pith.science/pith/J6VHBVZQ7JCASNI6CG352I6WF6/bundle.json","state":"https://pith.science/pith/J6VHBVZQ7JCASNI6CG352I6WF6/state.json","well_known_bundle":"https://pith.science/.well-known/pith/J6VHBVZQ7JCASNI6CG352I6WF6/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:J6VHBVZQ7JCASNI6CG352I6WF6","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":"aa00f64db9a6b6b71773a2aea7fd827867e5ec3fd65b7e65532ccd483aff4717","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-09T04:58:14Z","title_canon_sha256":"43be7874230e6864badccc0bfbe717a268a63fbb5d1669775c8edad857e18cc6"},"schema_version":"1.0","source":{"id":"2504.06577","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.06577","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"arxiv_version","alias_value":"2504.06577v1","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.06577","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"pith_short_12","alias_value":"J6VHBVZQ7JCA","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"pith_short_16","alias_value":"J6VHBVZQ7JCASNI6","created_at":"2026-07-05T10:46:29Z"},{"alias_kind":"pith_short_8","alias_value":"J6VHBVZQ","created_at":"2026-07-05T10:46:29Z"}],"graph_snapshots":[{"event_id":"sha256:192c8306f8d364cc22fd554d183a13925f3ff4fb16b93bd91df921d9bcc7adac","target":"graph","created_at":"2026-07-05T10:46:29Z","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/2504.06577/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In this paper, we show it is possible to bypass the safety guardrails of large language models (LLMs) through a humorous prompt including the unsafe request. In particular, our method does not edit the unsafe request and follows a fixed template -- it is simple to implement and does not need additional LLMs to craft prompts. Extensive experiments show the effectiveness of our method across different LLMs. We also show that both removing and adding more humor to our method can reduce its effectiveness -- excessive humor possibly distracts the LLM from fulfilling its unsafe request. Thus, we arg","authors_text":"Pedro Cisneros-Velarde","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-09T04:58:14Z","title":"Bypassing Safety Guardrails in LLMs Using Humor"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.06577","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":"sha256:6bb6b1d3bc92a24d07c4358a49c8249d86f052ab808c4a07a2bf58300fe75853","target":"record","created_at":"2026-07-05T10:46:29Z","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":"aa00f64db9a6b6b71773a2aea7fd827867e5ec3fd65b7e65532ccd483aff4717","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-09T04:58:14Z","title_canon_sha256":"43be7874230e6864badccc0bfbe717a268a63fbb5d1669775c8edad857e18cc6"},"schema_version":"1.0","source":{"id":"2504.06577","kind":"arxiv","version":1}},"canonical_sha256":"4faa70d730fa4409351e11b7dd23d62f8a6d86516039791d9dae5ff3d75d3092","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4faa70d730fa4409351e11b7dd23d62f8a6d86516039791d9dae5ff3d75d3092","first_computed_at":"2026-07-05T10:46:29.189624Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:46:29.189624Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"bNz+qCmWe8lrPL8MKiOG8iYU5/cazIwO9ZLy+Il7R1e+8f1T8ILb1qQsX+gLXjD8pOpKDS6OfCLjJCKzhZiKAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:46:29.190062Z","signed_message":"canonical_sha256_bytes"},"source_id":"2504.06577","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6bb6b1d3bc92a24d07c4358a49c8249d86f052ab808c4a07a2bf58300fe75853","sha256:192c8306f8d364cc22fd554d183a13925f3ff4fb16b93bd91df921d9bcc7adac"],"state_sha256":"b616199e2e03b54428d5c8bdcfedb730b2f07d1b8a53984ebedb1b24cc410eb2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5bWu7op6BDFXButM5TW+ELGZ3R1ulWhMluiHul/+rngmABPNfkdaAQjmU5FE4AI4FxTdkn5D06kwXcqJ0WhFCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-18T19:51:41.791356Z","bundle_sha256":"d3935ff345c9fe585f35aefa76761f57d0783414c73d3d44a60793e10d9da3e1"}}