{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:4HCJ4C7FZHUVFPGABOIBXNSUUI","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":"b53732f4a1830086c9f0d1d59abb7de9747d634d445637c59ad4870209e1db4a","cross_cats_sorted":["cs.AI","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-02-06T01:41:48Z","title_canon_sha256":"1085f3fc262ba35ad579cd861232adb577308936a79b69dce73d685fae15a19d"},"schema_version":"1.0","source":{"id":"2502.03708","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.03708","created_at":"2026-07-05T11:11:40Z"},{"alias_kind":"arxiv_version","alias_value":"2502.03708v2","created_at":"2026-07-05T11:11:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.03708","created_at":"2026-07-05T11:11:40Z"},{"alias_kind":"pith_short_12","alias_value":"4HCJ4C7FZHUV","created_at":"2026-07-05T11:11:40Z"},{"alias_kind":"pith_short_16","alias_value":"4HCJ4C7FZHUVFPGA","created_at":"2026-07-05T11:11:40Z"},{"alias_kind":"pith_short_8","alias_value":"4HCJ4C7F","created_at":"2026-07-05T11:11:40Z"}],"graph_snapshots":[{"event_id":"sha256:080393d47bf0b1e447c0edccb6fc98f96e9baeedd6975f44fff530b42adc6c9c","target":"graph","created_at":"2026-07-05T11:11:40Z","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/2502.03708/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Modern AI models contain much of human knowledge, yet understanding of their internal representation of this knowledge remains elusive. Characterizing the structure and properties of this representation will lead to improvements in model capabilities and development of effective safeguards. Building on recent advances in feature learning, we develop an effective, scalable approach for extracting linear representations of general concepts in large-scale AI models (language models, vision-language models, and reasoning models). We show how these representations enable model steering, through whi","authors_text":"Adityanarayanan Radhakrishnan, Daniel Beaglehole, Enric Boix-Adser\\`a, Mikhail Belkin","cross_cats":["cs.AI","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-02-06T01:41:48Z","title":"Toward universal steering and monitoring of AI models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.03708","kind":"arxiv","version":2},"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:bc0e89524c67ecbd28529fdbe229f029efbcbce0542bd4f6a525b59c9d5852ed","target":"record","created_at":"2026-07-05T11:11:40Z","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":"b53732f4a1830086c9f0d1d59abb7de9747d634d445637c59ad4870209e1db4a","cross_cats_sorted":["cs.AI","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-02-06T01:41:48Z","title_canon_sha256":"1085f3fc262ba35ad579cd861232adb577308936a79b69dce73d685fae15a19d"},"schema_version":"1.0","source":{"id":"2502.03708","kind":"arxiv","version":2}},"canonical_sha256":"e1c49e0be5c9e952bcc00b901bb654a22127261c0ec2ac70bcd8e2e118fbf3b6","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e1c49e0be5c9e952bcc00b901bb654a22127261c0ec2ac70bcd8e2e118fbf3b6","first_computed_at":"2026-07-05T11:11:40.681022Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:11:40.681022Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"EB/bmy1M9S4Fo4Uhok1nNKVAxlfbVuWT6FrIuOWgk0lUw7CHfLmapLliVANIMpKl5daz6Xyd06Buw5o1cRklBg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:11:40.681532Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.03708","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:bc0e89524c67ecbd28529fdbe229f029efbcbce0542bd4f6a525b59c9d5852ed","sha256:080393d47bf0b1e447c0edccb6fc98f96e9baeedd6975f44fff530b42adc6c9c"],"state_sha256":"67c5b1b1a26ebbf17438046bc35a4ceac49cce756e8c698cf7d8d312ac8892c1"}