{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:OAREBWB4D2SVRWWCAZOVP7MQMZ","short_pith_number":"pith:OAREBWB4","schema_version":"1.0","canonical_sha256":"702240d83c1ea558dac2065d57fd90665fc8ff4469365483a80122e4f66887b5","source":{"kind":"arxiv","id":"2503.03039","version":1},"attestation_state":"computed","paper":{"title":"LLM Misalignment via Adversarial RLHF Platforms","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ali Naseh, Erfan Entezami","submitted_at":"2025-03-04T22:38:54Z","abstract_excerpt":"Reinforcement learning has shown remarkable performance in aligning language models with human preferences, leading to the rise of attention towards developing RLHF platforms. These platforms enable users to fine-tune models without requiring any expertise in developing complex machine learning algorithms. While these platforms offer useful features such as reward modeling and RLHF fine-tuning, their security and reliability remain largely unexplored. Given the growing adoption of RLHF and open-source RLHF frameworks, we investigate the trustworthiness of these systems and their potential impa"},"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":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2503.03039","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-03-04T22:38:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"383b5e55a51644df19a9da5c41dd36c529a005cc3b5c1745ebf4f67e29e29227","abstract_canon_sha256":"b27d210f8d475e26d799af62b06b8d1431f6da97d474b176328f18e5a8e89ccf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:24:37.460492Z","signature_b64":"XH9bsPU2XsMRZQGPaaKBeP5aINERVE1curq9Z7XXix+AnzWYhugyhdU+No06rBU3OiT+JflKHrTeHYnEQos2BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"702240d83c1ea558dac2065d57fd90665fc8ff4469365483a80122e4f66887b5","last_reissued_at":"2026-07-05T10:24:37.459994Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:24:37.459994Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLM Misalignment via Adversarial RLHF Platforms","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ali Naseh, Erfan Entezami","submitted_at":"2025-03-04T22:38:54Z","abstract_excerpt":"Reinforcement learning has shown remarkable performance in aligning language models with human preferences, leading to the rise of attention towards developing RLHF platforms. These platforms enable users to fine-tune models without requiring any expertise in developing complex machine learning algorithms. While these platforms offer useful features such as reward modeling and RLHF fine-tuning, their security and reliability remain largely unexplored. Given the growing adoption of RLHF and open-source RLHF frameworks, we investigate the trustworthiness of these systems and their potential impa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.03039","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/2503.03039/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2503.03039","created_at":"2026-07-05T10:24:37.460051+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.03039v1","created_at":"2026-07-05T10:24:37.460051+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.03039","created_at":"2026-07-05T10:24:37.460051+00:00"},{"alias_kind":"pith_short_12","alias_value":"OAREBWB4D2SV","created_at":"2026-07-05T10:24:37.460051+00:00"},{"alias_kind":"pith_short_16","alias_value":"OAREBWB4D2SVRWWC","created_at":"2026-07-05T10:24:37.460051+00:00"},{"alias_kind":"pith_short_8","alias_value":"OAREBWB4","created_at":"2026-07-05T10:24:37.460051+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.02850","citing_title":"LLM Hypnosis: Exploiting User Feedback for Unauthorized Knowledge Injection to All Users","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16913","citing_title":"The Cognitive Penalty: Ablating System 1 and System 2 Reasoning in Edge-Native SLMs for Decentralized Consensus","ref_index":16,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ","json":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ.json","graph_json":"https://pith.science/api/pith-number/OAREBWB4D2SVRWWCAZOVP7MQMZ/graph.json","events_json":"https://pith.science/api/pith-number/OAREBWB4D2SVRWWCAZOVP7MQMZ/events.json","paper":"https://pith.science/paper/OAREBWB4"},"agent_actions":{"view_html":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ","download_json":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ.json","view_paper":"https://pith.science/paper/OAREBWB4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.03039&json=true","fetch_graph":"https://pith.science/api/pith-number/OAREBWB4D2SVRWWCAZOVP7MQMZ/graph.json","fetch_events":"https://pith.science/api/pith-number/OAREBWB4D2SVRWWCAZOVP7MQMZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ/action/storage_attestation","attest_author":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ/action/author_attestation","sign_citation":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ/action/citation_signature","submit_replication":"https://pith.science/pith/OAREBWB4D2SVRWWCAZOVP7MQMZ/action/replication_record"}},"created_at":"2026-07-05T10:24:37.460051+00:00","updated_at":"2026-07-05T10:24:37.460051+00:00"}