{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:646EFKCIIWOCX6PTUIL5OGAEQR","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":"c9af5a2efc49d71baea690d8516f6114c9fbc36c710e0d12663cc2fe29623263","cross_cats_sorted":["cs.CR","cs.LG"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-26T22:29:31Z","title_canon_sha256":"643729ae8ab8b5edce59204467db34360a37c62cd89453a04d397da02439851f"},"schema_version":"1.0","source":{"id":"2508.19461","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.19461","created_at":"2026-07-05T12:00:11Z"},{"alias_kind":"arxiv_version","alias_value":"2508.19461v1","created_at":"2026-07-05T12:00:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.19461","created_at":"2026-07-05T12:00:11Z"},{"alias_kind":"pith_short_12","alias_value":"646EFKCIIWOC","created_at":"2026-07-05T12:00:11Z"},{"alias_kind":"pith_short_16","alias_value":"646EFKCIIWOCX6PT","created_at":"2026-07-05T12:00:11Z"},{"alias_kind":"pith_short_8","alias_value":"646EFKCI","created_at":"2026-07-05T12:00:11Z"}],"graph_snapshots":[{"event_id":"sha256:58a25494eb31afde31fc228b200dcf3e8ab5c83af16b7a4fee5187f08faca26d","target":"graph","created_at":"2026-07-05T12:00:11Z","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/2508.19461/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We stress test monitoring systems for detecting covert misbehavior in autonomous LLM agents (e.g., secretly sharing private information). To this end, we systematize a monitor red teaming (MRT) workflow that incorporates: (1) varying levels of agent and monitor situational awareness; (2) distinct adversarial strategies to evade the monitor, such as prompt injection; and (3) two datasets and environments -- SHADE-Arena for tool-calling agents and our new CUA-SHADE-Arena, which extends TheAgentCompany, for computer-use agents. We run MRT on existing LLM monitor scaffoldings, which orchestrate LL","authors_text":"Ankit Aich, Chen Bo Calvin Zhang, Christina Q. Knight, Kevin Zhu, Neil Kale, Paula Rodriguez, Scale Red Team, Zifan Wang","cross_cats":["cs.CR","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-26T22:29:31Z","title":"Reliable Weak-to-Strong Monitoring of LLM Agents"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.19461","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":"sha256:0086fad6deb9a70c62711a68aae08fd51185d5aaa1ffc51f76a29517192b0a74","target":"record","created_at":"2026-07-05T12:00:11Z","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":"c9af5a2efc49d71baea690d8516f6114c9fbc36c710e0d12663cc2fe29623263","cross_cats_sorted":["cs.CR","cs.LG"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-26T22:29:31Z","title_canon_sha256":"643729ae8ab8b5edce59204467db34360a37c62cd89453a04d397da02439851f"},"schema_version":"1.0","source":{"id":"2508.19461","kind":"arxiv","version":1}},"canonical_sha256":"f73c42a848459c2bf9f3a217d71804847e3718fc288b383e30cb3c421104ddec","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f73c42a848459c2bf9f3a217d71804847e3718fc288b383e30cb3c421104ddec","first_computed_at":"2026-07-05T12:00:11.922738Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T12:00:11.922738Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"zm0GXbPaGGzrxpIsJiyCZMRNmwY8KcKsPCxwdGTI+0VxIxT7UpMYXxyOxwG69v3bmkdAf5r/9tgeYm/rAgPZAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T12:00:11.923254Z","signed_message":"canonical_sha256_bytes"},"source_id":"2508.19461","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0086fad6deb9a70c62711a68aae08fd51185d5aaa1ffc51f76a29517192b0a74","sha256:58a25494eb31afde31fc228b200dcf3e8ab5c83af16b7a4fee5187f08faca26d"],"state_sha256":"7d83a66ef94078398b7db7b51a7b657a16b1087eba42fe127157d5f5ab2cf501"}