{ "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": "2609.27620", "title": "InGuard: Towards Generalized Inner Guardrail for Safe Text-to-Image Generation", "summary": "InGuard replaces the conventional two-stage outer guardrail (prompt classifier plus post-hoc image classifier) with components operating on the T2I model's own representations: an embedding-based risk classifier, soft embedding modification (SAGE) to redirect risky prompts toward safe outputs, and a mid-denoising latent detector that halts generation early. The cost is training and validating three add-on components per deployed model, added pipeline complexity, and residual inference overhead, though base-model parameters remain untouched and flagged generations can exit after roughly half the denoising steps. It can fail on the mid-denoising latent estimate missing content that only becomes explicit in later steps, on graded 'risky' prompt handling degrading benign-prompt fidelity, and on the benchmark's reverse-generated prompts not covering the reader's actual threat distribution.", "rationale": "The method is deployable on a single 24 GB GPU since it adds small auxiliary modules without retraining the base model, but its safety gains must be validated against the reader's specific prompt distribution, models, and risk policies before production adoption.", "technique": "In-pipeline embedding guardrails for text-to-image safety", "family": "seguranca_e_guardrails", "recommendation": "testar", "published": "2026-09-23", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.27620", "independent_implementations": 0, "total_implementations": 1, "stars": 1, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "InGuard replaces the conventional two-stage outer guardrail (prompt classifier plus post-hoc image classifier) with components operating on the T2I model's own representations: an embedding-based risk classifier, soft embedding modification (SAGE) to redirect risky prompts toward safe outputs, and a mid-denoising latent detector that halts generation early. The cost is training and validating three add-on components per deployed model, added pipeline complexity, and residual inference overhead, though base-model parameters remain untouched and flagged generations can exit after roughly half the denoising steps. It can fail on the mid-denoising latent estimate missing content that only becomes explicit in later steps, on graded 'risky' prompt handling degrading benign-prompt fidelity, and on the benchmark's reverse-generated prompts not covering the reader's actual threat distribution.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.27620", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "97.9–98.8% safety rate across five open-weight T2I models, with 57.5–73.5% less benign disturbance, ~3.7x fewer parameters, and 50–55.6% of denoising steps skipped versus an outer guardrail baseline.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.27620", "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": "" }