{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VECMRUAT4I3YSWF3L3UCIDNP6K","short_pith_number":"pith:VECMRUAT","schema_version":"1.0","canonical_sha256":"a904c8d013e2378958bb5ee8240daff2b918faa7454a38aa349b58043f8cf029","source":{"kind":"arxiv","id":"2606.15980","version":2},"attestation_state":"computed","paper":{"title":"Do Activation Monitors Survive Model Updates? Benchmarking, Predicting, and Repairing Activation-Monitor Staleness","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Evan Duan","submitted_at":"2026-06-14T19:07:22Z","abstract_excerpt":"Activation monitors -- lightweight probes trained on a language model's internal representations -- are an increasingly common layer in deployment safety stacks. Deployed models however are rarely static: they are quantized, fine-tuned, adapted with LoRA, or served with merged adapters while the monitor remains frozen. We present the first systematic test of whether this implicit contract holds: whether activation monitors trained on a base model remain reliable after these routine model updates. Across multiple safety-relevant monitors, model depths, update families, and open-weight models, w"},"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":"2606.15980","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-14T19:07:22Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"7cd55cef3df07fe4b2cb0abbc5c325c3af50ca6648df5dd340bcdea48b2d2f24","abstract_canon_sha256":"58ef653de22daa54d7850f3db3f5fa7f1ad7b74104bfcd48165d8b9717277a47"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T01:12:09.933442Z","signature_b64":"/sb0sgT5NgDtdcaj5vY+fiZcQPdhxVEaHMNN37COutRyqUU79SYxngoCuL4rEFna5Q/lFPOFGEWc3QAgkWNTAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a904c8d013e2378958bb5ee8240daff2b918faa7454a38aa349b58043f8cf029","last_reissued_at":"2026-06-23T01:12:09.932970Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T01:12:09.932970Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Do Activation Monitors Survive Model Updates? Benchmarking, Predicting, and Repairing Activation-Monitor Staleness","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Evan Duan","submitted_at":"2026-06-14T19:07:22Z","abstract_excerpt":"Activation monitors -- lightweight probes trained on a language model's internal representations -- are an increasingly common layer in deployment safety stacks. Deployed models however are rarely static: they are quantized, fine-tuned, adapted with LoRA, or served with merged adapters while the monitor remains frozen. We present the first systematic test of whether this implicit contract holds: whether activation monitors trained on a base model remain reliable after these routine model updates. Across multiple safety-relevant monitors, model depths, update families, and open-weight models, w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.15980","kind":"arxiv","version":2},"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/2606.15980/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":"2606.15980","created_at":"2026-06-23T01:12:09.933047+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.15980v2","created_at":"2026-06-23T01:12:09.933047+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.15980","created_at":"2026-06-23T01:12:09.933047+00:00"},{"alias_kind":"pith_short_12","alias_value":"VECMRUAT4I3Y","created_at":"2026-06-23T01:12:09.933047+00:00"},{"alias_kind":"pith_short_16","alias_value":"VECMRUAT4I3YSWF3","created_at":"2026-06-23T01:12:09.933047+00:00"},{"alias_kind":"pith_short_8","alias_value":"VECMRUAT","created_at":"2026-06-23T01:12:09.933047+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/VECMRUAT4I3YSWF3L3UCIDNP6K","json":"https://pith.science/pith/VECMRUAT4I3YSWF3L3UCIDNP6K.json","graph_json":"https://pith.science/api/pith-number/VECMRUAT4I3YSWF3L3UCIDNP6K/graph.json","events_json":"https://pith.science/api/pith-number/VECMRUAT4I3YSWF3L3UCIDNP6K/events.json","paper":"https://pith.science/paper/VECMRUAT"},"agent_actions":{"view_html":"https://pith.science/pith/VECMRUAT4I3YSWF3L3UCIDNP6K","download_json":"https://pith.science/pith/VECMRUAT4I3YSWF3L3UCIDNP6K.json","view_paper":"https://pith.science/paper/VECMRUAT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.15980&json=true","fetch_graph":"https://pith.science/api/pith-number/VECMRUAT4I3YSWF3L3UCIDNP6K/graph.json","fetch_events":"https://pith.science/api/pith-number/VECMRUAT4I3YSWF3L3UCIDNP6K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VECMRUAT4I3YSWF3L3UCIDNP6K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VECMRUAT4I3YSWF3L3UCIDNP6K/action/storage_attestation","attest_author":"https://pith.science/pith/VECMRUAT4I3YSWF3L3UCIDNP6K/action/author_attestation","sign_citation":"https://pith.science/pith/VECMRUAT4I3YSWF3L3UCIDNP6K/action/citation_signature","submit_replication":"https://pith.science/pith/VECMRUAT4I3YSWF3L3UCIDNP6K/action/replication_record"}},"created_at":"2026-06-23T01:12:09.933047+00:00","updated_at":"2026-06-23T01:12:09.933047+00:00"}