{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.04067", "kind": "paper", "url": "https://arxiv.org/abs/2606.04067"}], "authors": ["Xinyue Huang", "Xiaochun Cao", "Wenyuan Yang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.04067", "source_url": "https://arxiv.org/abs/2606.04067"}, "categories": ["Biology & Drug Discovery", "Language & Knowledge", "cs.CR"], "description": {"en": "DelegateCI-Bench evaluates privacy-conscious query rewriting for LLM delegation, with 3,167 samples combining synthetic data across 20 task types, real user queries from WildChat, and a medical challenge set. Systems rewrite queries to suppress non-essential sensitive information."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.04067", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.04067", "title": "DelegateCI-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.04067", "modality": null, "name": "DelegateCI-Bench", "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_delegateci-bench_0d61203c", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "4d38b163577f7c17aeaad633791847b986f09fd08789f8441969cf604c493493", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.04067"}, "publisher": null, "released": "2026-06-02", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.04067", "source_url": "https://arxiv.org/abs/2606.04067"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-04067", "source": "claire_radar", "source_benchmark_id": "2606.04067", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Machine Generated", "applicationDomains": ["Biology & Drug Discovery"], "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/2606.04067", "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"}, "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": "96263337428be6b832d99ca1612b8c63f7071289fac80f14187161d10e542d53", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "DelegateCI-Bench evaluates privacy-conscious query rewriting for LLM delegation, with 3,167 samples combining synthetic data across 20 task types, real user queries from WildChat, and a medical challenge set. Systems rewrite queries to suppress non-essential sensitive information.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce DelegateCI-Bench, the first task based Contextual Integrity benchmark for privacy-conscious delegation, comprising 3,167 samples that combine high quality synthetic data spanning 11 tasks and 20 task types, WildChat based real user queries, and a medical challenge set with dense sensitive information."}, "familyId": "bmf_3a846b227d7c", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_delegateci-bench_0d61203c", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Pharma & Biotech"], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.04067", "paper": "https://arxiv.org/abs/2606.04067", "pdf": "https://arxiv.org/pdf/2606.04067", "project": null, "report": "https://arxiv.org/abs/2606.04067"}, "metrics": [], "motivation": "As LLMs become increasingly woven into everyday workflows, user queries sent to cloud hosted LLMs routinely mix task-essential content with task non-essential sensitive disclosures, yet type based PII redaction is context agnostic and may raise two issues: over disclosing untyped sensitive context and over removing answer bearing spans.", "name": "DelegateCI-Bench", "oneLine": "DelegateCI-Bench evaluates privacy-conscious query rewriting for LLM delegation, with 3,167 samples combining synthetic data across 20 task types, real user queries from WildChat, and a medical challenge set. Systems rewrite queries to suppress non-essential sensitive information.", "paperTitle": "Need to Know: Contextual-Integrity-Grounded Query Rewriting for Privacy-Conscious LLM Delegation", "primaryDomain": "Biology & Drug Discovery", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.04067", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-02", "source": {"authors": ["Xinyue Huang", "Xiaochun Cao", "Wenyuan Yang"], "categories": ["cs.CR", "cs.AI"], "id": "2606.04067", "title": "Need to Know: Contextual-Integrity-Grounded Query Rewriting for Privacy-Conscious LLM Delegation", "type": "arxiv", "url": "https://arxiv.org/abs/2606.04067"}, "sourceUpdatedAt": "2026-06-02T14:28:28Z", "topics": ["cs.CR"], "venueAttempts": [], "whyItMatters": "Addresses a gap in privacy benchmarks by focusing on task-based necessity rather than type-based PII redaction. Measures privacy-utility tradeoff in delegation, supporting safer LLM use."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
