{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "An agent reads external documents or tool responses that may contain malicious instructions.",
    "change": "Replay attacks and legitimate lookalikes in an isolated copy of your agent workflow.",
    "baseline": "Your existing checks with the same tasks, permissions, and tool responses.",
    "steps": [
      "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.",
      "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.",
      "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.",
      "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test."
    ],
    "measure": "Successful attacks missed, legitimate actions blocked, detection delay, and review cost.",
    "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation."
  },
  "builder_review": null,
  "arxiv_id": "2610.22930",
  "title": "Conformal Policy Selection: When to Deploy AI Decisions for Targeted Intervention",
  "summary": "It replaces ad hoc thresholding or average-effect criteria for deciding which individuals receive an AI-recommended treatment, substituting a multiple-testing procedure with weighted conformal p-values that bounds the false discovery rate among selected units. The cost is a held-out calibration set collected under the behavior policy, fitted propensity and outcome models, and added pipeline complexity rather than extra GPU or inference budget. It can fail under covariate shift beyond what the importance weights cover, when both nuisance models are misspecified simultaneously, or when calibration data are too scarce for the weighted conformal quantiles to be informative.",
  "rationale": "The method is computationally cheap and API-compatible, but it applies only to teams deploying consequential per-unit treatment or intervention decisions with logged behavior-policy data, so its value must be validated against the reader's actual workload rather than adopted unconditionally.",
  "technique": "Conformal policy selection with FDR control",
  "family": "seguranca_e_guardrails",
  "recommendation": "testar",
  "published": "2026-10-16",
  "as_of": "2026-10-19",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.22930",
  "independent_implementations": 0,
  "total_implementations": 1,
  "stars": 1573,
  "citations": null,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "It replaces ad hoc thresholding or average-effect criteria for deciding which individuals receive an AI-recommended treatment, substituting a multiple-testing procedure with weighted conformal p-values that bounds the false discovery rate among selected units. The cost is a held-out calibration set collected under the behavior policy, fitted propensity and outcome models, and added pipeline complexity rather than extra GPU or inference budget. It can fail under covariate shift beyond what the importance weights cover, when both nuisance models are misspecified simultaneously, or when calibration data are too scarce for the weighted conformal quantiles to be informative.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.22930",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "No multiplicative performance claim; the contribution is a statistical guarantee: finite-sample FDR control over the selected deployment set, with doubly robust asymptotic validity if either the outcome or propensity model is correctly specified, verified on synthetic and semi-synthetic EHR data.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.22930",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "exposure_map": [
    {
      "dimension": "quality",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compute",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "latency",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "operations",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compatibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "security",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "data_and_training",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "reproducibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "risks": [],
  "minimum_test": [],
  "open_questions": [],
  "independent_tests": [],
  "lab_reproductions": [],
  "report_available": false,
  "validation_tier": "unknown",
  "evidence_tier": "unknown",
  "infrastructure_basis": "unknown",
  "equations": [],
  "equations_status": "not_fetched",
  "core_kind": "",
  "equations_fetched_at": "",
  "equations_source_url": "",
  "equations_source_sha256": "",
  "exposures_reviewed_at": "",
  "exposures_source_sha256": ""
}