{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:NUZR3R2GAA26OHQIZEAP2XS4NX","short_pith_number":"pith:NUZR3R2G","schema_version":"1.0","canonical_sha256":"6d331dc7460035e71e08c900fd5e5c6dd30c429247b06b189df0eac05061c649","source":{"kind":"arxiv","id":"2605.12991","version":1},"attestation_state":"computed","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"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":true,"formal_links_present":false},"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"},"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"},"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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.12991","created_at":"2026-05-18T03:09:00.533804+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.12991v1","created_at":"2026-05-18T03:09:00.533804+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12991","created_at":"2026-05-18T03:09:00.533804+00:00"},{"alias_kind":"pith_short_12","alias_value":"NUZR3R2GAA26","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"NUZR3R2GAA26OHQI","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"NUZR3R2G","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX","json":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX.json","graph_json":"https://pith.science/api/pith-number/NUZR3R2GAA26OHQIZEAP2XS4NX/graph.json","events_json":"https://pith.science/api/pith-number/NUZR3R2GAA26OHQIZEAP2XS4NX/events.json","paper":"https://pith.science/paper/NUZR3R2G"},"agent_actions":{"view_html":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX","download_json":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX.json","view_paper":"https://pith.science/paper/NUZR3R2G","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.12991&json=true","fetch_graph":"https://pith.science/api/pith-number/NUZR3R2GAA26OHQIZEAP2XS4NX/graph.json","fetch_events":"https://pith.science/api/pith-number/NUZR3R2GAA26OHQIZEAP2XS4NX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/action/storage_attestation","attest_author":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/action/author_attestation","sign_citation":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/action/citation_signature","submit_replication":"https://pith.science/pith/NUZR3R2GAA26OHQIZEAP2XS4NX/action/replication_record"}},"created_at":"2026-05-18T03:09:00.533804+00:00","updated_at":"2026-05-18T03:09:00.533804+00:00"}