{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7X35R26IJOKXUK6ODWUMLD4P6J","short_pith_number":"pith:7X35R26I","schema_version":"1.0","canonical_sha256":"fdf7d8ebc84b957a2bce1da8c58f8ff275b121b1fb477ffa5412491c270f4ca7","source":{"kind":"arxiv","id":"2606.20408","version":1},"attestation_state":"computed","paper":{"title":"LLM agent safety, multi-turn red-teaming, jailbreak benchmarks, adversarial robustness, safety-critical systems","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Bokyeong Kim, Dasol Choi, Hanwool Lee, Haon Park, Seung Geun Kim","submitted_at":"2026-06-18T15:57:53Z","abstract_excerpt":"Large language model (LLM) agents are increasingly proposed as supervisory components for safety-critical systems, yet their robustness under sustained, adaptive adversarial pressure remains poorly characterized. We present NRT-Bench, a benchmark for multi-turn red-teaming of LLM agents acting as operators of a safety-critical system, instantiated in a simulated nuclear power plant control room. A five-role operator team, each backed by a configurable LLM, runs a plant governed by six critical safety functions (CSFs), while adversaries inject messages over four channels in bounded multi-turn s"},"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":"2606.20408","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-06-18T15:57:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"570114d7bbbdbd5f4043ad224f26f611479b491469f76792e4a59ab9aa90ee9d","abstract_canon_sha256":"378118fcf708d4ff07624e1fb27ce547ccba7bad6abcf9fddc7a05f66a8a28ea"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:13:11.780656Z","signature_b64":"j4oV6AVPVDk6s3yT/rEb3B/92iXpVCoyOhH5oex82B0ZYV1FSpkMTveR/8DW3u+MsWzAPJt5pVmDzsMzoM/TAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fdf7d8ebc84b957a2bce1da8c58f8ff275b121b1fb477ffa5412491c270f4ca7","last_reissued_at":"2026-06-19T16:13:11.780280Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:13:11.780280Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLM agent safety, multi-turn red-teaming, jailbreak benchmarks, adversarial robustness, safety-critical systems","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Bokyeong Kim, Dasol Choi, Hanwool Lee, Haon Park, Seung Geun Kim","submitted_at":"2026-06-18T15:57:53Z","abstract_excerpt":"Large language model (LLM) agents are increasingly proposed as supervisory components for safety-critical systems, yet their robustness under sustained, adaptive adversarial pressure remains poorly characterized. We present NRT-Bench, a benchmark for multi-turn red-teaming of LLM agents acting as operators of a safety-critical system, instantiated in a simulated nuclear power plant control room. A five-role operator team, each backed by a configurable LLM, runs a plant governed by six critical safety functions (CSFs), while adversaries inject messages over four channels in bounded multi-turn s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.20408","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/2606.20408/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":"2606.20408","created_at":"2026-06-19T16:13:11.780343+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.20408v1","created_at":"2026-06-19T16:13:11.780343+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.20408","created_at":"2026-06-19T16:13:11.780343+00:00"},{"alias_kind":"pith_short_12","alias_value":"7X35R26IJOKX","created_at":"2026-06-19T16:13:11.780343+00:00"},{"alias_kind":"pith_short_16","alias_value":"7X35R26IJOKXUK6O","created_at":"2026-06-19T16:13:11.780343+00:00"},{"alias_kind":"pith_short_8","alias_value":"7X35R26I","created_at":"2026-06-19T16:13:11.780343+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/7X35R26IJOKXUK6ODWUMLD4P6J","json":"https://pith.science/pith/7X35R26IJOKXUK6ODWUMLD4P6J.json","graph_json":"https://pith.science/api/pith-number/7X35R26IJOKXUK6ODWUMLD4P6J/graph.json","events_json":"https://pith.science/api/pith-number/7X35R26IJOKXUK6ODWUMLD4P6J/events.json","paper":"https://pith.science/paper/7X35R26I"},"agent_actions":{"view_html":"https://pith.science/pith/7X35R26IJOKXUK6ODWUMLD4P6J","download_json":"https://pith.science/pith/7X35R26IJOKXUK6ODWUMLD4P6J.json","view_paper":"https://pith.science/paper/7X35R26I","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.20408&json=true","fetch_graph":"https://pith.science/api/pith-number/7X35R26IJOKXUK6ODWUMLD4P6J/graph.json","fetch_events":"https://pith.science/api/pith-number/7X35R26IJOKXUK6ODWUMLD4P6J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7X35R26IJOKXUK6ODWUMLD4P6J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7X35R26IJOKXUK6ODWUMLD4P6J/action/storage_attestation","attest_author":"https://pith.science/pith/7X35R26IJOKXUK6ODWUMLD4P6J/action/author_attestation","sign_citation":"https://pith.science/pith/7X35R26IJOKXUK6ODWUMLD4P6J/action/citation_signature","submit_replication":"https://pith.science/pith/7X35R26IJOKXUK6ODWUMLD4P6J/action/replication_record"}},"created_at":"2026-06-19T16:13:11.780343+00:00","updated_at":"2026-06-19T16:13:11.780343+00:00"}