All capabilities · Product, business & communication

Apply responsible-AI and data-protection basics

Bias, transparency, consent, DPDP Act/GDPR basics, model documentation.

~6 focused hoursbeginner
Explore 3 tools for this project
What employers mean

You should be able to…

  1. Apply DPDP Act (India) basics — consent, purpose limitation, data minimization — to an AI feature
  2. Check a model or dataset for obvious demographic bias before launch
  3. Write clear consent/disclosure language when a user is interacting with an AI system
  4. Document a model's known limitations for internal and external audiences
  5. Identify when a use case needs a human-in-the-loop for legal or ethical reasons
  6. Explain the difference between DPDP Act and GDPR obligations for an Indian company serving global users

Needs first: Explain how LLMs work and where they fail

Learn — free, link-checked

The few resources that matter

Tools for practice

Choose a tool for the job

Start with one tool for each part of your project. You don’t need to learn them all.

Go to the practice brief

3 tools to explore

Google Docs

Plan & explain

Write a rubric, project story or decision brief that others can review and comment on.

Google Sheets

Data · Plan & explain

Build a scoring sheet, clean a small dataset or make assumptions visible in a simple model.

Practices & references

  • Model and data documentation
  • Bias-testing checklists
  • DPDP and GDPR reference guidance
Practice

Responsible-AI checklist and model card for a resume-screening assistant

Take a hypothetical feature that screens resumes for a recruiting team, and write it up against the actual law rather than a summary of it: MeitY publishes the Digital Personal Data Protection Act 2023 as a free PDF. Produce (1) a candidate-facing consent and data-handling note citing the sections it rests on, (2) a bias-check plan naming specific demographic attributes and the method for testing each, and (3) a one-page model card stating known limitations and the points where a human must review.

Start from

The Digital Personal Data Protection Act 2023 — MeitY's published PDF, the sections you cite for consent, purpose limitation and data minimisation

Milestones
  1. Read the DPDP Act and pull the sections that bind a resume-screening use case · ~1.5h
  2. Write the candidate consent and data-handling note against those sections · ~1.5h
  3. Write the bias-check plan: three attributes, a method for each · ~0.5h
  4. Write the model card and reconcile all three documents against each other · ~0.5h
Done when
  • Consent note cites specific DPDP Act obligations (purpose limitation, consent, data minimization)
  • Bias-check plan names at least 3 attributes to test and a method for each
  • Model card states at least 2 known limitations and where a human must review before a decision
  • All three documents are consistent with each other (no contradictions on data handling)
Prove it

Evidence a recruiter can check

  • A consent note that cites DPDP Act sections by number, not 'in line with data-protection law'
  • A bias-check plan naming three attributes with the specific method and data slice used to test each
  • A model card listing two known limitations and the exact decision points where a human has to review
  • A reconciliation pass showing all three documents agree on what data is kept and for how long
Signal it

Wrote the responsible-AI pack for a resume-screening feature — a DPDP-cited consent note, a three-attribute bias-check plan, and a model card fixing where a human must sign off before a rejection.

Interview

Questions you'll get asked

  1. What does the DPDP Act require when an AI feature processes personal data?
  2. How would you check whether a hiring-screening model is biased before launch?
  3. What disclosure would you require when a user is chatting with a bot instead of a human?
  4. Tell me about a use case that needed a human-in-the-loop for compliance reasons, not just accuracy.
  5. How do DPDP Act and GDPR obligations differ for an Indian company with EU customers?
  6. What would you put in a model documentation card for a customer-facing LLM feature?