AI assurance resources
AI assurance evidence library.
Practical workbooks, guides, and research for evaluating AI systems, evidence, human oversight, vendor risk, model risk, and governance.
Start with a downloadable review tool or browse the library by topic.
01 / Start here
Choose by task.
- Review an AI system
- AI Assurance Evidence Review Kit v2.2.1General workbook for system context, risks, controls, evidence, human oversight, gaps, and follow-up. View the Review Kit.
- Review an AML AI workflow
- AML AI Assurance Evidence Kit v1.1Specialist workbook connecting claims, evidence, tests, human oversight, findings and decisions for AI-supported AML monitoring. View the AML kit.
- Research a topic
- Guides, references, and analysisBrowse assurance, risk, oversight, vendor, model, framework, LLM/RAG, and AML resources. Browse topics.
02 / Workbooks & toolkits
Current releases.
Download the workbook or open its resource page for methodology, scope, and release details.
General AI assurance · v2.2.1
AI Assurance Evidence Review Kit
A fourteen-worksheet workbook for reviewing AI claims, risks, controls, evidence, tests and human intervention, then assigning findings and recording decisions. Start Here, Field Guide and a connected Worked Example are included.
Best fit
AI governance, technology risk, internal audit, model risk, and control reviews centered on a defined AI system or use case.
Core worksheets
- Start Here and Review Brief
- Claims and Risks & Controls
- Evidence and Tests
- Human Oversight
- Findings and Decisions
- Review Summary
- Field Guide, Review Library, Worked Example and Sources & Method
AML assurance · v1.1
AML AI Assurance Evidence Kit
A sixteen-worksheet Excel workbook for reviewing AI-supported AML monitoring and alert prioritization. Connect claims, evidence, tests, human oversight, findings and decisions, with specialist workflow and model/data-change records.
Best fit
AML assurance, compliance review, model governance, internal audit, and teams reviewing how AI-supported monitoring and alert decisions are documented, escalated, and evidenced.
Included in the workbook
- Start Here and Review Brief
- Claims, Risks & Controls, Evidence and Tests
- Human Oversight, Findings and Decisions
- AML Workflow and Model & Data Events
- Review Summary
- Field Guide, 24 AML review prompts, Worked Example and Sources & Method
03 / Guides & references
Browse by topic.
Reference pages for common AI assurance, governance, and risk questions.
-
AI assurance
Evidence, assurance claims, review methods, and bounded conclusions.
Foundations
-
AI risk domains
Failure modes, controls, ownership, and evidence across major AI risk areas.
Risk
-
Human oversight
Authority, intervention, override, escalation, rationale, and follow-through.
Oversight
-
Vendor AI risk
Supplier claims, responsibilities, evidence access, monitoring, and change.
Third party
-
Model risk support
Documentation, performance, limitations, validation inputs, and monitoring.
Model risk
-
AI frameworks, standards & regulations
Primary-source directory for governance, regulation, security, audit, and assurance resources.
Reference
-
LLM / RAG risk
Prompt, retrieval, output, access, monitoring, dependencies, and human review.
GenAI
-
AML AI human oversight evidence
Analyst rationale, reviewer roles, escalation, override, intervention, and records.
AML
04 / Research & analysis
Research and analysis.
Selected InfoSecured analysis on AI governance, financial AI, and explainability.
-
Why Financial AI Governance Needs a Global Risk Framework
Regulatory fragmentation, systemic AI risk, and interoperable governance in financial services.
Financial AI
-
Generative AI in Finance: 2025 Playbook From the Swiss Bankers Association
Governance, legal, compliance, and implementation considerations for GenAI in banking.
GenAI
-
Explainable AI in Finance
Explainability, transparency, compliance, and review expectations for financial AI.
Explainability
05 / About
About the Evidence Library.
The Evidence Library is InfoSecured’s catalog of public AI assurance workbooks, guides, references, and research. Each downloadable resource has a dedicated page for its methodology, scope, version, and release details.