{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.16723", "kind": "paper", "url": "https://arxiv.org/abs/2606.16723"}, {"id": "gh:rohithreddybc/agentfairbench", "kind": "repo", "url": "https://github.com/rohithreddybc/AgentFairBench"}], "authors": ["Triveni Morla", "Rohith Reddy Bellibatlu", "Manpreet Singh", "Manmeet Singh Kapoor"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.16723", "source_url": "https://arxiv.org/abs/2606.16723"}, "categories": ["General AI", "Language & Knowledge", "cs.AI"], "description": {"en": "AgentFairBench evaluates demographic disparity in the actions of LLM agents across hiring, lending, and medical triage. It uses synthetic, demographic-neutral profiles in counterfactual matched sets varying name-coded race/gender. Metrics include counterfactual flip rate, mean absolute score difference, action-rate disparity, and tool-invocation disparity, with bootstrap confidence intervals and FDR control."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.16723", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.16723", "title": "AgentFairBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.16723", "modality": null, "name": "AgentFairBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_agentfairbench_1a8330ea", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "25184616560f19fef9df646602733b6ff169f389f81a2c1d9421ae217ced0087", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.16723"}, "publisher": null, "released": "2026-06-15", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.16723", "source_url": "https://arxiv.org/abs/2606.16723"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-16723", "source": "claire_radar", "source_benchmark_id": "2606.16723", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Machine Generated", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/rohithreddybc/AgentFairBench", "githubScope": "benchmark_repo", "githubStars": 2, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 1, "hfPaperUrl": "https://huggingface.co/papers/2606.16723", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-10-01", "state": "stale"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Original Synthetic", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "867d877722466981e8b6a05208f9e754896d1d37100e8809e931de311d2e2aa0", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "AgentFairBench evaluates demographic disparity in the actions of LLM agents across hiring, lending, and medical triage. It uses synthetic, demographic-neutral profiles in counterfactual matched sets varying name-coded race/gender. Metrics include counterfactual flip rate, mean absolute score difference, action-rate disparity, and tool-invocation disparity, with bootstrap confidence intervals and FDR control.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce AgentFairBench, a cheap, reproducible, multi-domain benchmark for demographic disparity in the actions of LLM agents."}, "familyId": "bmf_806b46eeb591", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "linked-from-report", "linkedUrl": "https://github.com/rohithreddybc/AgentFairBench", "url": "https://arxiv.org/abs/2606.16723"}], "scope": "benchmark_repo", "url": "https://github.com/rohithreddybc/AgentFairBench", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_agentfairbench_1a8330ea", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/rohithreddybc/AgentFairBench", "data": null, "hfPaper": "https://huggingface.co/papers/2606.16723", "paper": "https://arxiv.org/abs/2606.16723", "pdf": "https://arxiv.org/pdf/2606.16723", "project": null, "report": "https://arxiv.org/abs/2606.16723"}, "metrics": [], "motivation": "Large language model (LLM) agents increasingly take actions (screening applicants, recommending credit, triaging patients), yet fairness for LLMs is still measured by grading answers.", "name": "AgentFairBench", "oneLine": "AgentFairBench evaluates demographic disparity in the actions of LLM agents across hiring, lending, and medical triage. It uses synthetic, demographic-neutral profiles in counterfactual matched sets varying name-coded race/gender. Metrics include counterfactual flip rate, mean absolute score difference, action-rate disparity, and tool-invocation disparity, with bootstrap confidence intervals and FDR control.", "paperTitle": "AgentFairBench: Do LLM Agents Discriminate When They Act?", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.16723", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-15", "source": {"authors": ["Triveni Morla", "Rohith Reddy Bellibaltu", "Manpreet Singh", "Manmeet Singh Kapoor"], "categories": ["cs.AI"], "id": "2606.16723", "title": "AgentFairBench: Do LLM Agents Discriminate When They Act?", "type": "arxiv", "url": "https://arxiv.org/abs/2606.16723"}, "sourceUpdatedAt": "2026-06-15T13:50:26Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "Existing fairness evaluations grade answers, not actions. AgentFairBench addresses the gap by measuring disparity in consequential agent decisions. Its low cost and reproducible harness provide a practical path for screening models for action-level bias before deployment."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
