Skip to main content
ToolDix logoToolDixFind, compare, and use tools
Sign In
Browse 20 learning paths

AI learning path

Responsible AI & Security

Turn responsible-AI principles into risk registers, threat models, evaluation gates, incident plans, and accountable operating practices.

For Builders, security teams, product owners, and governance leaders

8 original lessons on ToolDix, roughly 144 minutes end to end

Start the course

Course outline

  1. 1Build an AI Risk Register People Actually UseTurn vague AI concerns into scenario-level entries with named owners, detection signals, and regression tests, so the register drives engineering work instead of sitting in a compliance folder.Intermediate20 min
  2. 2Threat-Model an LLM ApplicationSort every component into trust zones, find the four transitions where untrusted content silently gains authority, and derive abuse cases from assets so the model produces regression tests rather than a reading list.Advanced22 min
  3. 3Prompt Injection: Why Filtering Cannot Fix ItUnderstand why a language model cannot separate instructions from data, how indirect injection reaches you through your own retrieval pipeline, and which controls actually bound the damage.Advanced18 min
  4. 4Sensitive Information Disclosure in LLM SystemsMap the six routes by which private data escapes an AI feature, apply the data minimization ladder, and enforce tenant isolation in the retrieval query rather than in the answer.Advanced17 min
  5. 5Excessive Agency and Least Privilege for AgentsSeparate the three dimensions of agency, fix the confused deputy problem with identity propagation, and place approval at the action tier instead of the session.Advanced17 min
  6. 6Improper Output HandlingTreat model output as untrusted input, map every sink it can reach, and apply the encoding each destination requires instead of one generic sanitize step.Advanced16 min
  7. 7Red-Team an AI SystemRun adversarial testing as a repeatable loop with scoped objectives, technique coverage, impact-based severity, and findings that become permanent regression tests.Advanced17 min
  8. 8AI Supply Chain SecurityInventory the six layers you inherit when shipping an AI feature, pin and mirror model artifacts like packages, and understand why poisoning evades aggregate metrics.Advanced17 min

Optional, after the lessons

Further reading from primary sources

Published by other organizations and kept here for reference. The lessons above are the ToolDix course; these are where to go once you want the vendor's own documentation.

7 resources

Introduction to Responsible AI by Google for Developers
CourseBeginner

Introduction to Responsible AI

Google for Developers

A structured introduction to fairness, accountability, safety, and privacy considerations when developing and scaling AI systems.

Time
Short learning modules
fairnessaccountabilityprivacy
Read the ToolDix guide
AI Risk Management Framework by NIST
Classic readingIntermediate

AI Risk Management Framework

NIST

A voluntary framework organized around Govern, Map, Measure, and Manage for operationalizing trustworthy and responsible AI risk management.

Time
Framework and playbook
NIST AI RMFrisk managementgovernance
Read the ToolDix guide
Generative AI Profile by NIST
Classic readingAdvanced

Generative AI Profile

NIST

A companion profile that applies the AI RMF to generative-AI risks, actions, measurement needs, and governance considerations.

Time
Companion risk profile
generative AIrisk profilecontrols
Read the ToolDix guide
OWASP Top 10 for LLM Applications by OWASP GenAI Security Project
Hands-on labIntermediate

OWASP Top 10 for LLM Applications

OWASP GenAI Security Project

A practitioner reference for common LLM application risks such as prompt injection, sensitive-data disclosure, excessive agency, and insecure output handling.

Time
Threat-by-threat guide
prompt injectionexcessive agencythreat modeling
Read the ToolDix guide
Microsoft Responsible AI Standard by Microsoft
Classic readingAdvanced

Microsoft Responsible AI Standard

Microsoft

A public overview of Microsoft's responsible-AI principles, governance approach, and resources for putting accountability into product development.

Time
Standard and implementation guidance
responsible AIstandardsaccountability
Read the ToolDix guide
Secure AI Framework (SAIF) by Google
Classic readingIntermediate

Secure AI Framework (SAIF)

Google

A security framework for mapping AI risks, extending established controls, automating defenses, and adapting protection to model, data, infrastructure, and application layers.

Time
Framework · risk map · controls
AI securityrisk assessmentcontrols
Read the ToolDix guide
MITRE ATLAS by MITRE
Classic readingAdvanced

MITRE ATLAS

MITRE

A structured knowledge base of adversary tactics, techniques, case studies, and mitigations for machine-learning and generative-AI systems.

Time
Living knowledge base
adversary tacticsAI red teamingmitigations
Read the ToolDix guide