1. Foundations of Responsible AI
This course introduces the core principles of establishing ethical oversight, regulatory compliance, and governance frameworks across the AI lifecycle. This course provides a practical overview of AI fundamentals, ethical principles, and compliance requirements within Indian and global regulatory landscapes. Through applied workflows, participants learn to navigate AI capabilities and limits, set up intake procedures and inventories, establish risk tiering workflows, and implement the operational governance structures needed to build transparent and accountable AI applications.
AI Basics & Need for Governance
- What AI, ML, and Generative AI Are
- Key Capabilities and Common Limits
- Real‑world Impacts of AI Systems
- Why Governance and Oversight Are Needed
Ethics & Core Principles
- Ethics vs Compliance in AI
- Fairness, Accountability, Transparency
- Safety, Reliability, Inclusiveness
- Privacy and Security as Ethical Duties
Laws, Regulations & Standards
- Global AI Frameworks & EU AI Act Overview
- AI Guidelines (MeitY)
- DPDP Act & IT Act
- Sectoral Regulators in India
- Internal Policies & Standards
AI Lifecycle & Governance Process
- Use‑case Intake and AI Inventory
- Risk Tiering and Approval Workflows
- Roles, Responsibilities, and Decision Rights
- Documentation and Audit Trails for Models
2. AI Risk Management & Safety Engineering
This course introduces the essential frameworks and methodologies for identifying, evaluating, and mitigating risks across Machine Learning and Generative AI deployments. This course provides a practical deep dive into data quality, algorithmic bias, operational safety, transparency, privacy, and LLM-specific vulnerabilities such as hallucinations and prompt injections. Through applied evaluation techniques, participants learn to conduct threat modeling, perform red-teaming, implement technical guardrails, manage real-time incidents, and establish continuous auditing processes to ensure secure, reliable, and compliant AI systems.
Data & Information Risk
- Data Quality, Lineage, and Provenance Issues
- Consent, Purpose Limitation & Minimization Gaps
- Sensitive and Personal Data Handling Risks
- Third‑Party Data Use and Licensing Risks
Fairness, Bias & Discrimination
- Sources of Bias in Datasets and Models
- Bias Leading to Unfair Outcomes
- Fairness Metrics & Evaluation Approaches
- Techniques to Reduce and Monitor Bias
Safety, Reliability & Operations
- AI Failure Modes and Error Patterns
- Hallucinations & Unsafe Behavior
- Testing, Red‑teaming & Stress Testing
- Guardrails, Fallbacks & Human Review
Transparency & Explainability
- Clarity on System Purpose and Limits
- Black‑box Decisions and User Mistrust
- High‑Level Explainability Methods
- User Notices, Labels, and Appeal Options
Privacy, Security & IP Risk
- Threat Modeling for AI Systems
- Data Leakage via Prompts, Logs & Outputs
- Unauthorized Access & Model Misuse
- IP, Copyright & Licensing of Data
Generative AI & LLM-Specific Risk
- Hallucinations & Fabricated Information
- Prompt Injection & Jailbreak Attempts
- Quality, Factuality & Brand‑Safety Risks
- Vendor & Model Selection Due Diligence
Incidents & Audits
- Incident Categories & Severity Levels
- Triage, Response, and Remediation Steps
- Internal and External Audit Processes
- Learning from Incidents & Updating Controls
