{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:NUZR3R2GAA26OHQIZEAP2XS4NX","short_pith_number":"pith:NUZR3R2G","canonical_record":{"source":{"id":"2605.12991","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T04:45:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7eed6309f5d7ee2b84ef9f6e1749195760118488965057c30381d575a094d572","abstract_canon_sha256":"fd6bf71118822b5767fbb0c2cfba5854e4f9bcc37b7883da0aa25e25dbf3c215"},"schema_version":"1.0"},"canonical_sha256":"6d331dc7460035e71e08c900fd5e5c6dd30c429247b06b189df0eac05061c649","source":{"kind":"arxiv","id":"2605.12991","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12991","created_at":"2026-05-18T03:09:00Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12991v1","created_at":"2026-05-18T03:09:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12991","created_at":"2026-05-18T03:09:00Z"},{"alias_kind":"pith_short_12","alias_value":"NUZR3R2GAA26","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"NUZR3R2GAA26OHQI","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"NUZR3R2G","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:NUZR3R2GAA26OHQIZEAP2XS4NX","target":"record","payload":{"canonical_record":{"source":{"id":"2605.12991","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T04:45:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7eed6309f5d7ee2b84ef9f6e1749195760118488965057c30381d575a094d572","abstract_canon_sha256":"fd6bf71118822b5767fbb0c2cfba5854e4f9bcc37b7883da0aa25e25dbf3c215"},"schema_version":"1.0"},"canonical_sha256":"6d331dc7460035e71e08c900fd5e5c6dd30c429247b06b189df0eac05061c649","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:09:00.534216Z","signature_b64":"tfkmpKXkeQzT6qhG3WE/pIbi7+mCs8ULdtWRatm6O49d/kDfYKrVQRcr+Ra/FnOBqt8aCdUxIEQ54J77lBBACg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6d331dc7460035e71e08c900fd5e5c6dd30c429247b06b189df0eac05061c649","last_reissued_at":"2026-05-18T03:09:00.533726Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:09:00.533726Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.12991","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":"2026-05-18T03:09:00Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LorABG9Ed+zd1FO6Jn7vskFfpUSMg6HmA3o/soatFUwC/mTqRmB1coK6YM3jQGt8LLnnWEcmIlImbkiPM0TWBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T23:41:52.752997Z"},"content_sha256":"327b592555198f85e9b3319055f96c17fe26655ba12b77e0bfab6766e943a888","schema_version":"1.0","event_id":"sha256:327b592555198f85e9b3319055f96c17fe26655ba12b77e0bfab6766e943a888"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:NUZR3R2GAA26OHQIZEAP2XS4NX","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Not Just RLHF: Why Alignment Alone Won't Fix Multi-Agent Sycophancy","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Pretrained base models exhibit the same or higher yield to simulated peer disagreement as their RLHF-tuned counterparts, localizing the issue to mid-layer attention rather than alignment.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Adarsh Kumarappan, Ananya Mujoo","submitted_at":"2026-05-13T04:45:08Z","abstract_excerpt":"LLM-based multi-agent pipelines flip from correct to incorrect answers under simulated peer disagreement at rates we term yield, a vulnerability widely attributed to RLHF-induced sycophancy. We test this attribution across four model families and find it largely wrong: pretrained base models exhibit the same substitution pattern as their Instruct variants, averaging higher yield than Instruct. Using activation patching, we localize the corruption to a narrow mid-layer window where attention carries the causal weight and MLP contribution is negligible; patching above this window restores 96% of"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"pretrained base models exhibit the same substitution pattern as their Instruct variants, averaging higher yield than Instruct. Using activation patching, we localize the corruption to a narrow mid-layer window where attention carries the causal weight.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the simulated peer disagreement in the experimental setup accurately captures the dynamics of real multi-agent LLM pipelines and that yield directly measures sycophancy rather than other forms of uncertainty or context sensitivity.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Pretrained base models exhibit higher yield to peer disagreement than RLHF instruct variants, with the effect localized to mid-layer attention and mitigated by structured dissent rather than prompt defenses.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Pretrained base models exhibit the same or higher yield to simulated peer disagreement as their RLHF-tuned counterparts, localizing the issue to mid-layer attention rather than alignment.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"264ffab35af6fb2c2553ed9582eaa566d94dc447d2931374fe7a2ae4a418656a"},"source":{"id":"2605.12991","kind":"arxiv","version":1},"verdict":{"id":"e6b72d73-65ad-440b-8517-967df60d18d2","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:00:18.935475Z","strongest_claim":"pretrained base models exhibit the same substitution pattern as their Instruct variants, averaging higher yield than Instruct. Using activation patching, we localize the corruption to a narrow mid-layer window where attention carries the causal weight.","one_line_summary":"Pretrained base models exhibit higher yield to peer disagreement than RLHF instruct variants, with the effect localized to mid-layer attention and mitigated by structured dissent rather than prompt defenses.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the simulated peer disagreement in the experimental setup accurately captures the dynamics of real multi-agent LLM pipelines and that yield directly measures sycophancy rather than other forms of uncertainty or context sensitivity.","pith_extraction_headline":"Pretrained base models exhibit the same or higher yield to simulated peer disagreement as their RLHF-tuned counterparts, localizing the issue to mid-layer attention rather than alignment."},"references":{"count":39,"sample":[{"doi":"","year":null,"title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone","work_id":"feef9556-a016-493c-abd2-0c97a23a7ebf","ref_index":1,"cited_arxiv_id":"2404.14219","is_internal_anchor":true},{"doi":"","year":null,"title":"Constitutional AI: Harmlessness from AI Feedback","work_id":"faaaa4e0-2676-4fac-a0b4-99aef10d2095","ref_index":2,"cited_arxiv_id":"2212.08073","is_internal_anchor":true},{"doi":"","year":null,"title":"Small Language Models are the Future of Agentic AI","work_id":"ba0f0305-4a51-48fd-a13f-201439a18f9e","ref_index":3,"cited_arxiv_id":"2506.02153","is_internal_anchor":true},{"doi":"","year":null,"title":"Eliciting Latent Predictions from Transformers with the Tuned Lens","work_id":"a127314f-7424-488f-b6d7-8214650c420f","ref_index":4,"cited_arxiv_id":"2303.08112","is_internal_anchor":true},{"doi":"","year":null,"title":"Measuring Progress on Scalable Oversight for Large Language Models","work_id":"e7f92eb1-2050-4e60-bc27-82c94d7694c5","ref_index":5,"cited_arxiv_id":"2211.03540","is_internal_anchor":true}],"resolved_work":39,"snapshot_sha256":"2337783ab501485f3e5793d2792bbd98a697eb19239becea7314993b670e9c7b","internal_anchors":26},"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":"e6b72d73-65ad-440b-8517-967df60d18d2"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:09:00Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"2HTiRjIP9CvNxHsWMxgCeDXNKU8D2IornZls77ALjLsgu6qnW+mzMDZiVX7fCefm8ezb/05gjY7Gb08W38WhDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T23:41:52.753986Z"},"content_sha256":"4f4ccdbda782916f236c66897ab3702c345f7869bca46d8ab14fb4fa5f156ab1","schema_version":"1.0","event_id":"sha256:4f4ccdbda782916f236c66897ab3702c345f7869bca46d8ab14fb4fa5f156ab1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/bundle.json","state_url":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/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":"2026-05-25T23:41:52Z","links":{"resolver":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX","bundle":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/bundle.json","state":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/state.json","well_known_bundle":"https://pith.science/.well-known/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:NUZR3R2GAA26OHQIZEAP2XS4NX","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":"fd6bf71118822b5767fbb0c2cfba5854e4f9bcc37b7883da0aa25e25dbf3c215","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T04:45:08Z","title_canon_sha256":"7eed6309f5d7ee2b84ef9f6e1749195760118488965057c30381d575a094d572"},"schema_version":"1.0","source":{"id":"2605.12991","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12991","created_at":"2026-05-18T03:09:00Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12991v1","created_at":"2026-05-18T03:09:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12991","created_at":"2026-05-18T03:09:00Z"},{"alias_kind":"pith_short_12","alias_value":"NUZR3R2GAA26","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"NUZR3R2GAA26OHQI","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"NUZR3R2G","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:4f4ccdbda782916f236c66897ab3702c345f7869bca46d8ab14fb4fa5f156ab1","target":"graph","created_at":"2026-05-18T03:09:00Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"pretrained base models exhibit the same substitution pattern as their Instruct variants, averaging higher yield than Instruct. Using activation patching, we localize the corruption to a narrow mid-layer window where attention carries the causal weight."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the simulated peer disagreement in the experimental setup accurately captures the dynamics of real multi-agent LLM pipelines and that yield directly measures sycophancy rather than other forms of uncertainty or context sensitivity."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Pretrained base models exhibit higher yield to peer disagreement than RLHF instruct variants, with the effect localized to mid-layer attention and mitigated by structured dissent rather than prompt defenses."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Pretrained base models exhibit the same or higher yield to simulated peer disagreement as their RLHF-tuned counterparts, localizing the issue to mid-layer attention rather than alignment."}],"snapshot_sha256":"264ffab35af6fb2c2553ed9582eaa566d94dc447d2931374fe7a2ae4a418656a"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"LLM-based multi-agent pipelines flip from correct to incorrect answers under simulated peer disagreement at rates we term yield, a vulnerability widely attributed to RLHF-induced sycophancy. We test this attribution across four model families and find it largely wrong: pretrained base models exhibit the same substitution pattern as their Instruct variants, averaging higher yield than Instruct. Using activation patching, we localize the corruption to a narrow mid-layer window where attention carries the causal weight and MLP contribution is negligible; patching above this window restores 96% of","authors_text":"Adarsh Kumarappan, Ananya Mujoo","cross_cats":["cs.AI"],"headline":"Pretrained base models exhibit the same or higher yield to simulated peer disagreement as their RLHF-tuned counterparts, localizing the issue to mid-layer attention rather than alignment.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T04:45:08Z","title":"Not Just RLHF: Why Alignment Alone Won't Fix Multi-Agent Sycophancy"},"references":{"count":39,"internal_anchors":26,"resolved_work":39,"sample":[{"cited_arxiv_id":"2404.14219","doi":"","is_internal_anchor":true,"ref_index":1,"title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone","work_id":"feef9556-a016-493c-abd2-0c97a23a7ebf","year":null},{"cited_arxiv_id":"2212.08073","doi":"","is_internal_anchor":true,"ref_index":2,"title":"Constitutional AI: Harmlessness from AI Feedback","work_id":"faaaa4e0-2676-4fac-a0b4-99aef10d2095","year":null},{"cited_arxiv_id":"2506.02153","doi":"","is_internal_anchor":true,"ref_index":3,"title":"Small Language Models are the Future of Agentic AI","work_id":"ba0f0305-4a51-48fd-a13f-201439a18f9e","year":null},{"cited_arxiv_id":"2303.08112","doi":"","is_internal_anchor":true,"ref_index":4,"title":"Eliciting Latent Predictions from Transformers with the Tuned Lens","work_id":"a127314f-7424-488f-b6d7-8214650c420f","year":null},{"cited_arxiv_id":"2211.03540","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Measuring Progress on Scalable Oversight for Large Language Models","work_id":"e7f92eb1-2050-4e60-bc27-82c94d7694c5","year":null}],"snapshot_sha256":"2337783ab501485f3e5793d2792bbd98a697eb19239becea7314993b670e9c7b"},"source":{"id":"2605.12991","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:00:18.935475Z","id":"e6b72d73-65ad-440b-8517-967df60d18d2","model_set":{"reader":"grok-4.3"},"one_line_summary":"Pretrained base models exhibit higher yield to peer disagreement than RLHF instruct variants, with the effect localized to mid-layer attention and mitigated by structured dissent rather than prompt defenses.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Pretrained base models exhibit the same or higher yield to simulated peer disagreement as their RLHF-tuned counterparts, localizing the issue to mid-layer attention rather than alignment.","strongest_claim":"pretrained base models exhibit the same substitution pattern as their Instruct variants, averaging higher yield than Instruct. Using activation patching, we localize the corruption to a narrow mid-layer window where attention carries the causal weight.","weakest_assumption":"That the simulated peer disagreement in the experimental setup accurately captures the dynamics of real multi-agent LLM pipelines and that yield directly measures sycophancy rather than other forms of uncertainty or context sensitivity."}},"verdict_id":"e6b72d73-65ad-440b-8517-967df60d18d2"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:327b592555198f85e9b3319055f96c17fe26655ba12b77e0bfab6766e943a888","target":"record","created_at":"2026-05-18T03:09:00Z","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":"fd6bf71118822b5767fbb0c2cfba5854e4f9bcc37b7883da0aa25e25dbf3c215","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T04:45:08Z","title_canon_sha256":"7eed6309f5d7ee2b84ef9f6e1749195760118488965057c30381d575a094d572"},"schema_version":"1.0","source":{"id":"2605.12991","kind":"arxiv","version":1}},"canonical_sha256":"6d331dc7460035e71e08c900fd5e5c6dd30c429247b06b189df0eac05061c649","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6d331dc7460035e71e08c900fd5e5c6dd30c429247b06b189df0eac05061c649","first_computed_at":"2026-05-18T03:09:00.533726Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:00.533726Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"tfkmpKXkeQzT6qhG3WE/pIbi7+mCs8ULdtWRatm6O49d/kDfYKrVQRcr+Ra/FnOBqt8aCdUxIEQ54J77lBBACg==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:00.534216Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12991","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:327b592555198f85e9b3319055f96c17fe26655ba12b77e0bfab6766e943a888","sha256:4f4ccdbda782916f236c66897ab3702c345f7869bca46d8ab14fb4fa5f156ab1"],"state_sha256":"f9b9a00b8645dd9d19f4a8ecca5c79f56e832bbe5b8c1a421de227c6182e12ca"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lwVFegIADrxwCw9JLVoscJVcjE7zKn0jk+pta8MPs1ClPkA0kFm7EvX++yFWqM73ysu6O3jDMTsRvs6V59sPBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-25T23:41:52.758237Z","bundle_sha256":"d91efab917015f85c03e054fc3043764931e5219154f8602e0ee4a994972f114"}}