{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DEDY7WTKLLKERMTHQ3PPIMMWFJ","short_pith_number":"pith:DEDY7WTK","schema_version":"1.0","canonical_sha256":"19078fda6a5ad448b26786def431962a4f50cb4c38fcbc35bd1002750067a1f6","source":{"kind":"arxiv","id":"2605.26526","version":1},"attestation_state":"computed","paper":{"title":"Open-Weight LLM Fine-Tuning Defenses are Susceptible to Simple Attacks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Chhavi Yadav, Kevin Kuo, Virginia Smith","submitted_at":"2026-05-26T04:18:42Z","abstract_excerpt":"Recent defenses for safeguarding open-weight large language models (LLMs) are intended to prevent adversarial usage. Underlying these defenses is an assumption that new harmful behavior is learned through fine-tuning rather than elicited by jailbreaking the model. Yet, pretrained LLMs already encode substantial harmful knowledge across many domains, which raises an important question: can an adversary jailbreak safeguarded models, to achieve harmful usage without fine-tuning at all? In this paper, we show that open-weight safeguards are susceptible to simpler strategies that, despite being wel"},"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":"2605.26526","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-26T04:18:42Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"069c2fb54a149384c42ed89d93af31f3f84a100aed6c5ea0e8ec91b205b756a7","abstract_canon_sha256":"272ed8c1ad7f63381684889c13d4199512c509670f7fbcb855ce3f60fc6705b8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:05:25.270063Z","signature_b64":"8D0mnCG7Cbj+4+sXPHSjT6jooO/2vOPp5HRiC0lEl9yofx7kuemqfEc702CPPUaz2UtGrYw5WVqY8qXLaO8+Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"19078fda6a5ad448b26786def431962a4f50cb4c38fcbc35bd1002750067a1f6","last_reissued_at":"2026-05-27T01:05:25.269535Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:05:25.269535Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Open-Weight LLM Fine-Tuning Defenses are Susceptible to Simple Attacks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Chhavi Yadav, Kevin Kuo, Virginia Smith","submitted_at":"2026-05-26T04:18:42Z","abstract_excerpt":"Recent defenses for safeguarding open-weight large language models (LLMs) are intended to prevent adversarial usage. Underlying these defenses is an assumption that new harmful behavior is learned through fine-tuning rather than elicited by jailbreaking the model. Yet, pretrained LLMs already encode substantial harmful knowledge across many domains, which raises an important question: can an adversary jailbreak safeguarded models, to achieve harmful usage without fine-tuning at all? In this paper, we show that open-weight safeguards are susceptible to simpler strategies that, despite being wel"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26526","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/2605.26526/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":"2605.26526","created_at":"2026-05-27T01:05:25.269611+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26526v1","created_at":"2026-05-27T01:05:25.269611+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26526","created_at":"2026-05-27T01:05:25.269611+00:00"},{"alias_kind":"pith_short_12","alias_value":"DEDY7WTKLLKE","created_at":"2026-05-27T01:05:25.269611+00:00"},{"alias_kind":"pith_short_16","alias_value":"DEDY7WTKLLKERMTH","created_at":"2026-05-27T01:05:25.269611+00:00"},{"alias_kind":"pith_short_8","alias_value":"DEDY7WTK","created_at":"2026-05-27T01:05:25.269611+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/DEDY7WTKLLKERMTHQ3PPIMMWFJ","json":"https://pith.science/pith/DEDY7WTKLLKERMTHQ3PPIMMWFJ.json","graph_json":"https://pith.science/api/pith-number/DEDY7WTKLLKERMTHQ3PPIMMWFJ/graph.json","events_json":"https://pith.science/api/pith-number/DEDY7WTKLLKERMTHQ3PPIMMWFJ/events.json","paper":"https://pith.science/paper/DEDY7WTK"},"agent_actions":{"view_html":"https://pith.science/pith/DEDY7WTKLLKERMTHQ3PPIMMWFJ","download_json":"https://pith.science/pith/DEDY7WTKLLKERMTHQ3PPIMMWFJ.json","view_paper":"https://pith.science/paper/DEDY7WTK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26526&json=true","fetch_graph":"https://pith.science/api/pith-number/DEDY7WTKLLKERMTHQ3PPIMMWFJ/graph.json","fetch_events":"https://pith.science/api/pith-number/DEDY7WTKLLKERMTHQ3PPIMMWFJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DEDY7WTKLLKERMTHQ3PPIMMWFJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DEDY7WTKLLKERMTHQ3PPIMMWFJ/action/storage_attestation","attest_author":"https://pith.science/pith/DEDY7WTKLLKERMTHQ3PPIMMWFJ/action/author_attestation","sign_citation":"https://pith.science/pith/DEDY7WTKLLKERMTHQ3PPIMMWFJ/action/citation_signature","submit_replication":"https://pith.science/pith/DEDY7WTKLLKERMTHQ3PPIMMWFJ/action/replication_record"}},"created_at":"2026-05-27T01:05:25.269611+00:00","updated_at":"2026-05-27T01:05:25.269611+00:00"}