{ "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": "2609.16403", "title": "Implementing a White-Box Undetectable Backdoor for Random Fourier Features", "summary": "The paper implements, in numpy and scipy, a cryptographically undetectable backdoor for Random Fourier Features models, replacing informal assumptions about white-box auditability with a concrete demonstration that full weight inspection cannot distinguish backdoored from clean models. It does not replace any production technique; the cost to the reader is awareness that weight-level auditing of RFF-based models offers no security guarantee, and replicating the work requires deriving sampling machinery the original construction left unspecified. What can fail: the finding applies to the RFF setting specifically, the underlying lattice hardness reduction was not reproduced, and the negative indistinguishability result is empirical within the tested sparsity range rather than a new proof.", "rationale": "This is a security-realizability result with no deployable performance benefit; it informs threat modeling when accepting third-party RFF models but requires no production changes on constrained infrastructure.", "technique": "White-box CLWE backdoor for Random Fourier Features", "family": "seguranca_e_guardrails", "recommendation": "observar", "published": "2026-09-14", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.16403", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": 0, "claims": [ { "claim_id": "claim-01", "statement": "The paper implements, in numpy and scipy, a cryptographically undetectable backdoor for Random Fourier Features models, replacing informal assumptions about white-box auditability with a concrete demonstration that full weight inspection cannot distinguish backdoored from clean models. It does not replace any production technique; the cost to the reader is awareness that weight-level auditing of RFF-based models offers no security guarantee, and replicating the work requires deriving sampling machinery the original construction left unspecified. What can fail: the finding applies to the RFF setting specifically, the underlying lattice hardness reduction was not reproduced, and the negative indistinguishability result is empirical within the tested sparsity range rather than a new proof.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.16403", "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": [ { "anchor": "S4.E1", "label": "(1)", "section": "4 Implementation", "role": "proposed_method", "latex": "p(t)\\;\\propto\\;e^{-t^{2}/2}\\sum_{k\\in\\mathbb{Z}}\\exp\\!\\left(-\\frac{(k+s-\\gamma t)^{2}}{2\\beta^{2}}\\right),", "mathml": "p(t)∝e−t2/2∑k∈ℤexp(−(k+s−γ​t)22​β2),", "context": "The distribution G​Pd​(bk) factors through the homogeneous CLWE density of Bruna et al. [2]. Restricted to the projection t=⟨g,u⟩ onto the secret direction u=ω/‖ω‖, this density is, by their Definition 2.19, a discrete-Gaussian-weighted mixture of Gaussians:", "headline": "", "explanation": "", "objective": "", "symbols": {}, "evidence": "", "evidence_url": "", "source_url": "" } ], "equations_status": "selected", "core_kind": "formula", "equations_fetched_at": "2026-10-14", "equations_source_url": "", "equations_source_sha256": "", "exposures_reviewed_at": "", "exposures_source_sha256": "" }