{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:NFHHVA54S2MFYIOQN52AVMLTSO","merge_version":"pith-open-graph-merge-v1","event_count":3,"valid_event_count":3,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"8c660850af84e57d0a829b65724145a97b744bf9a579475a4a49c5cc0e6f967a","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-07-28T20:06:31Z","title_canon_sha256":"0dbd7d296b394372dd5913c33962b184106bff2f35a8bf3e864ad8c1d004b337"},"schema_version":"1.0","source":{"id":"2607.26228","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.26228","created_at":"2026-07-30T01:17:45Z"},{"alias_kind":"arxiv_version","alias_value":"2607.26228v1","created_at":"2026-07-30T01:17:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.26228","created_at":"2026-07-30T01:17:45Z"},{"alias_kind":"pith_short_12","alias_value":"NFHHVA54S2MF","created_at":"2026-07-30T01:17:45Z"},{"alias_kind":"pith_short_16","alias_value":"NFHHVA54S2MFYIOQ","created_at":"2026-07-30T01:17:45Z"},{"alias_kind":"pith_short_8","alias_value":"NFHHVA54","created_at":"2026-07-30T01:17:45Z"}],"graph_snapshots":[{"event_id":"sha256:1f20243edd847da6f711a745b443b748dd2ff93fea24d60900306f63893ce3c6","target":"graph","created_at":"2026-07-30T01:17:45Z","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/2607.26228/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Instruction hierarchies are a core safety assumption of language model deployment: higher priority inputs, such as system prompts, should override conflicting lower priority inputs from users or tools. Yet frontier LLMs often violate this hierarchy. We introduce V-Steer, a training-free inference time method that restores privileged influence by editing cached value vectors at prompt positions. Using direct logit attribution on the first next token prediction, V-Steer identifies heads where lower priority spans dominate privileged ones, then boosts privileged spans and suppresses conflicting l","authors_text":"Han Zhao, Julia Hockenmaier, Sewoong Lee, Siqi Zeng","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-07-28T20:06:31Z","title":"Steering Instruction Hierarchies at Inference Time"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.26228","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":"sha256:742c7efcb0ce1cc653149e251b618b446a163d10524d7e993ad3306cd06b0bef","target":"record","created_at":"2026-07-30T01:17:45Z","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":"8c660850af84e57d0a829b65724145a97b744bf9a579475a4a49c5cc0e6f967a","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-07-28T20:06:31Z","title_canon_sha256":"0dbd7d296b394372dd5913c33962b184106bff2f35a8bf3e864ad8c1d004b337"},"schema_version":"1.0","source":{"id":"2607.26228","kind":"arxiv","version":1}},"canonical_sha256":"694e7a83bc96985c21d06f740ab17393a997d15ca9035098168ed52e3db7fdf2","receipt":{"builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"694e7a83bc96985c21d06f740ab17393a997d15ca9035098168ed52e3db7fdf2","first_computed_at":"2026-07-30T01:17:45.772981Z","kind":"pith_receipt","last_reissued_at":"2026-07-30T01:17:45.772981Z","receipt_version":"0.3","signature_status":"unsigned_v0"},"source_id":"2607.26228","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:742c7efcb0ce1cc653149e251b618b446a163d10524d7e993ad3306cd06b0bef","sha256:1f20243edd847da6f711a745b443b748dd2ff93fea24d60900306f63893ce3c6","sha256:c2303515b408abe2aa4901b1172bce6fcf9f739866e713791d17137d40bd1e39"],"state_sha256":"4794a199baa40b08275398e64b699172bd900d2156611487b713a670db9075b8"}