{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2609.09754", "kind": "paper", "url": "https://arxiv.org/abs/2609.09754"}], "authors": ["Yujin Zhou", "Mingxuan Zheng", "Chuxue Cao", "Huang Yidan", "Jiale Chen", "Yike Guo", "Sirui Han"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2609.09754", "source_url": "https://arxiv.org/abs/2609.09754"}, "categories": ["General AI", "Language & Knowledge", "cs.AI"], "description": {"en": "LexAgentHallu is a hierarchical benchmark for profiling hallucinations in tool-augmented legal agents, containing 3414 instances across 17 legal categories and 6 task types, with dual-layer hallucination taxonomy and fine-grained metrics."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2609.09754", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2609.09754", "title": "LexAgentHallu"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2609.09754", "modality": null, "name": "LexAgentHallu", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "false", "origin_record_id": "bm_lexagenthallu_8032f3b2", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "aa01e9e90027c19bf412c4c1947434795e3c5d21083851408f51756f76ce0cf4", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-name-audit-deferred", "reviewed_at": "2026-09-11T00:52:53.785364Z", "source_url": "https://arxiv.org/abs/2609.09754"}, "publisher": null, "released": "2026-09-09", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2609.09754", "source_url": "https://arxiv.org/abs/2609.09754"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2609-09754", "source": "claire_radar", "source_benchmark_id": "2609.09754", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2609.09754", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "The legal domain niche and detailed hallucination taxonomy may attract specialist attention, but lack of released artifacts limits immediate engagement.", "score": 45}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-11T00:52:53.785364Z", "inputHash": "e10db0945aaa352227fada7ab0dafc3447d677cf8027b9eb22e9f59dcd4220eb", "model": "deepseek-v4-pro"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "we introduce LexAgentHallu, a legal agentic hallucination benchmark", "canonicalNameSource": "abstract", "decisionReason": "The absence of any released code, dataset, or public submission path makes the benchmark inaccessible for independent verification; defer until artifacts are available.", "evaluationEvidence": {"reuse": "the benchmark is introduced for evaluation, implying public availability through the arxiv report", "scoring": "evaluates to what extent and how legal agents fail along multi-step trajectories", "task": "legal agentic hallucination benchmark designed to evaluate to what extent and how legal agents fail along multi-step trajectories"}, "model": "deepseek-v4-pro", "reviewedAt": "2026-09-11T00:52:53.785364Z", "state": "ai-name-audit-deferred"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "LexAgentHallu is a hierarchical benchmark for profiling hallucinations in tool-augmented legal agents, containing 3414 instances across 17 legal categories and 6 task types, with dual-layer hallucination taxonomy and fine-grained metrics.", "displayEligible": false, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "To address these limitations, we introduce LexAgentHallu, a legal agentic hallucination benchmark designed to evaluate to what extent and how legal agents fail along multi-step trajectories."}, "familyId": "bmf_6ebe3c2ca064", "firstSeenAt": "2026-09-10", "heat": null, "id": "bm_lexagenthallu_8032f3b2", "indexedAt": "2026-09-10T00:47:26.889471Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2609.09754", "pdf": "https://arxiv.org/pdf/2609.09754", "project": null, "report": "https://arxiv.org/abs/2609.09754"}, "metrics": [], "motivation": "As large language models are increasingly deployed as tool-augmented legal agents, they introduce agentic hallucinations where tool-call and reasoning errors cascade into fabricated holdings and miscited authority.", "name": "LexAgentHallu", "oneLine": "LexAgentHallu is a hierarchical benchmark for profiling hallucinations in tool-augmented legal agents, containing 3414 instances across 17 legal categories and 6 task types, with dual-layer hallucination taxonomy and fine-grained metrics.", "paperTitle": "LexAgentHallu: A Hierarchical Benchmark for Profiling Hallucinations in Legal Agents", "primaryDomain": "General AI", "publication": {"evidence": "EMNLP 2026 Main", "evidenceLevel": "strong-author-metadata", "evidenceUrl": "https://arxiv.org/abs/2609.09754", "source": "arxiv-journal-reference", "status": "publication_reported", "venue": "EMNLP 2026 Main", "verifiedAt": "2026-09-10T00:47:26.889471Z"}, "publications": [{"evidence": [{"level": "strong-author-metadata", "observedAt": "2026-09-10T00:47:26.889471Z", "rawValue": "EMNLP 2026 Main", "sourceType": "arxiv-journal-reference", "sourceUrl": "https://arxiv.org/abs/2609.09754"}], "publicationStatus": "published", "venueName": "EMNLP 2026 Main"}], "ranking": {"30d": {"confidence": "Low", "coverage": 0.0, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": 30, "historyDate": "2026-09-05", "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v13", "rank": null, "score": null}, "90d": {"confidence": "Low", "coverage": 0.0, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v13", "rank": null, "score": null}}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-09-09", "source": {"authors": ["Yujin Zhou", "Mingxuan Zheng", "Chuxue Cao", "Huang Yidan", "Jiale Chen", "Yike Guo", "Sirui Han"], "categories": ["cs.AI"], "id": "2609.09754", "title": "LexAgentHallu: A Hierarchical Benchmark for Profiling Hallucinations in Legal Agents", "type": "arxiv", "url": "https://arxiv.org/abs/2609.09754"}, "sourceUpdatedAt": "2026-09-09T05:52:53Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "It addresses the lack of diagnostic benchmarks for legal agent hallucinations, aiming to reveal failure modes invisible to outcome-level evaluation."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
