KoraSafe™ maps the four SR 11-7 requirements (model inventory and tiering, independent validation evidence, governance and board reporting, ongoing monitoring) to a single platform your model risk team and your examiner can both read from. Two pieces are still gaps; we name them below.
SR 11-7 is the Federal Reserve's supervisory letter on model risk management. SR 15-18 supplements it for complex and judgmental models, including modern ML and AI. OCC Bulletin 2011-12 carries the same requirements for OCC-supervised institutions. FFIEC joint guidance builds on all three for the broader US banking population. KoraSafe™ maps to the requirements common to all four documents, cited by section.
Every AI system is registered with owner, autonomy tier, decomposed risk score, lifecycle state, and regulatory tier before it reaches production. Five-tier autonomy maps to SR 11-7 materiality tiering. The registry is the inventory SR 11-7 §II requires, in the format your model risk team can hand to an examiner without reformatting.
Hallucination detection covers conceptual soundness for generative models (groundedness against authoritative sources). Fairness Watchdog covers outcomes analysis (disparate-impact monitoring across protected attributes). Behavioral Drift Detector covers ongoing distribution-shift validation. Anomaly Killer covers effective challenge with statistical outlier detection. Each guardian writes findings mapped to SR 11-7 §IV by section, not flagged generically.
Policy authoring, approval workflows, lifecycle transitions, and dry-runs against staging produce the paper trail §V requires. Governance index publishes a board-readable score quarterly with drilldown to the underlying findings, model versions, and policy events. Senior management gets the quarterly model risk view regulators expect.
Drift Detector and Anomaly Killer run continuously against per-system baselines. Findings link back to the model, the time window, and the SR 11-7 monitoring section, so your next exam starts from a documented record. Risk score history tracks performance degradation per system over time.
Every finding traces back to the regulation section, the model version, the policy that fired, and the reviewer who closed it. SR 11-7 audit-package endpoint generates SR 11-7-keyed evidence bundles with vendor attestations, model risk approvals, and fairness findings, pre-organized for examiner review.
SR 11-7 doesn't run alone in a model risk program. The financial services pack carries FFIEC joint-agency guidance, NAIC Model Bulletin for insurance carriers, ECOA / Reg B for fair-lending evidence, NYDFS Part 500 for state cybersecurity sign-off. Same audit chain, same evidence format, same examiner-ready output.
Most SR 11-7 components are live. Two are not. Surfacing them here rather than in the AC of a sales call.
Guardians produce findings mapped to SR 11-7 §IV. KoraSafe™ does not yet emit a bundled "SR 11-7 validation report" document per model, in the format examiners expect for a model validation deliverable. Today, your model risk team assembles the validation report from the guardian findings, the risk score history, and the audit-chain evidence package. A native validation-report artifact is on the roadmap.
Policy lifecycle transitions exist with re-approval triggers. The equivalent state machine for model-level changes (retrain, version bump, use-case extension, decommission) is not yet wired with automatic re-validation triggers. Today, model changes flow through your existing change-management process; KoraSafe™ captures the event but doesn't enforce re-validation gates. Roadmap.
Inventory, validation evidence, governance, monitoring, audit chain
Financial services sector pack with SR 11-7 mapping
Validation report artifact + model change lifecycle
Adjacent vendors handle pieces of SR 11-7, none of them name the four requirements together. Honest read of current public materials:
Corrections welcome at Contact-us@korasafe.ai. Public competitive positioning shifts faster than this page; we date-stamp the comparison so readers know what window we surveyed.
If your model risk team has an SR 11-7 review on the calendar, KoraSafe™ is built for that exam window. Start your free trial for onboarding, or jump to the financial services pack to see how SR 11-7 sits inside the broader FinServ coverage.