Week 5 · Lesson 2 · 45-55 minutes
Protect Privacy and Minimize Data
Learn what counts as personal data, why consent matters, and how data minimization keeps an AI feature working while collecting far less about people.
- Materials
- This lesson in a web browser · Paper and pencil, or a notes app
- This lesson
- In progress
Auto-saves on this device
What you'll be able to do
- Define privacy and personal data and give examples of each.
- Explain what consent means and why it must be informed.
- Apply data minimization to keep only the data a feature truly needs.
- Redesign a described feature so it collects less personal data.
Think about it
A free flashlight app asks to see your contacts, your exact location, and your photos before it will turn on the light. A flashlight only needs to control the light. Why might an app ask for far more data than its job requires?
Hold on to your answer — by the end you will be able to redesign a feature so it stops over-collecting.
Make a prediction
Predict: for a weather app to show today's forecast, which is enough — your city, or your exact street address and full location history?
Vocabulary
See the full course vocabulary- Privacy
- A person's right to control who can see and use information about them.
- Personal data
- Any information that is about a particular person, such as a name, address, exact location, birthday, photo, or messages.
- Consent
- A person clearly agreeing to let their data be collected or used, after being told plainly what will happen with it.
- Data minimization
- Collecting and keeping only the data a feature truly needs to work, and no more.
Privacy and personal data: what needs protecting
Privacy is your right to control who can see and use information about you. Personal data is any information that is about a particular person — a name, a home address, an exact location, a birthday, a photo of a face, or private messages.
Not all data is personal. 'It rained on Tuesday' is not about any one person. But 'Ana was at the park at 4:00' is personal data, because it tells someone about a specific person. AI features often run on personal data, which is exactly why privacy matters here.
- Personal data: full name, home address, exact GPS location, face photo.
- Usually not personal: the current weather, a public bus schedule.
- Sensitive personal data: health information, private messages.
Consent means an informed yes, not a hidden one
Consent is a person clearly agreeing to let their data be collected or used. For consent to be real, it must be informed: the person is told, in plain language, what data is taken and what it is used for, and they can say no.
Consent buried in fine print, pre-checked boxes, or a confusing pop-up is not real consent, because the person never truly understood or freely chose. Good design asks clearly and makes 'no' an easy, safe option.
- Clear consent: 'Share your location so we can show nearby stops? Yes / No.'
- Not real consent: a long agreement that hides data sharing in paragraph 40.
- Not real consent: an app that will not work unless you allow unrelated tracking.
Data minimization: collect only what the job needs
Data minimization is a simple, powerful rule: collect and keep only the data a feature actually needs to work. If a feature can do its job with less, it should. Data you never collect cannot be leaked, misused, or seen by the wrong people.
To minimize, ask for each piece of data, 'Does the feature truly need this to work?' A weather forecast needs your general area, not your exact address or your entire location history. Cutting the extra data keeps the feature useful while lowering the risk to people.
- A forecast that uses your city instead of your precise GPS trail.
- A step counter that stores today's total, not everywhere you walked.
- A sign-up that asks for a nickname instead of your legal full name.
Less data, same job: minimization in practice
Minimizing does not mean breaking the feature. Most features can be redesigned to need less by using a broader or shorter-lived version of the data. Instead of an exact location, use a general area. Instead of keeping data forever, delete it once the task is done.
This is a design choice made by people. When you build or judge an AI feature, you can push for the version that collects the least personal data while still doing its job — that is respecting privacy by design, not by accident.
- Turn 'exact address' into 'city' for a weather feature.
- Turn 'keep forever' into 'delete after the trip ends' for a map route.
- Turn 'real name and birthday' into 'nickname only' for a game leaderboard.
Worked example
Minimizing a school bus-tracker feature
- Name the feature's job. Show a student when their bus is about five minutes away so they can head to the stop.
- List what a careless version might collect. The student's exact GPS location all day, their full name, their home address, and their whole location history.
- Ask the minimization question for each item. To alert 'bus is near your stop,' the feature needs the one stop the student uses — not a full day of tracking.
- Redesign with less. Let the student pick their stop from a list. Store only that stop, and only compare it to the bus location while the bus is running.
- Check consent. Ask clearly, 'Alert you when the bus nears your stop? Yes / No,' and make sure the student and family understand and can decline.
Takeaway: Start from the feature's real job, then keep only the smallest data that still does that job — and ask for it with clear consent.
A before-and-after comparison of a bus-tracker feature. Before (over-collecting): stores the student's exact location all day, full legal name, home address, and complete location history kept forever. After (minimized): stores only the one bus stop the student selected, checked against the bus location only while the bus is running, and deleted at the end of the day. The 'after' version does the same job — alerting the student when the bus is near — while holding far less personal data.
Same job, far less personal data: the minimized design lowers the risk to the student.
| Data requested | Needed for a local forecast? | Keep it? |
|---|---|---|
| General area or city | Yes — forecast is local | Keep |
| Exact GPS address | No — city is enough | Drop |
| Full location history | No — only current area matters | Drop |
| Contacts list | No — unrelated to weather | Drop |
A table testing each piece of data a weather feature might request against whether it is truly needed. General area or city: needed, because the forecast is local. Exact GPS address: not needed, the city is enough. Full location history: not needed, only current area matters. Contacts list: not needed, unrelated to weather. The rule of thumb is to keep only rows marked 'needed'.
Activity
Privacy and Data Minimization Lab
For several fictional apps, classify each data field as required, helpful, unnecessary, or too sensitive — then compare with guidance on safer alternatives, consent, and how long to keep data.
- Pick an app scenario. For each data field, decide: required, helpful, unnecessary, or too sensitive.
- Aim to collect only what the app truly needs — data minimization.
- Reveal the guidance to compare, and see safer alternatives, consent, and how long to keep each field.
These are made-up apps. Don't enter any of your own personal details — just classify the fields.
- First name
- Exact birthday
- Age range
- School name
- Exact location
- Broad region (city/area)
- Voice recording
- Photo
- Favorite subject
- Anonymous activity choices
- Device identifier
- Contact list
- Accessibility preference
Check your understanding
Answer these to check that you understand personal data, consent, and data minimization.
Question 1. Which of these count as personal data? Select all that apply.
Select all that apply.
Question 2. Which choice best follows data minimization?
A quiz-game app needs a name to show on the score list. It offers three options for what to collect from each player.
Question 3. Decide if the statement is true or false.
Hiding an agreement to share data deep in fine print, where users are unlikely to read it, still counts as real, informed consent.
0 of 3 answered
Try it yourself
Minimize a feature you use
Pick a common app feature and redesign it to collect the least personal data while still working.
- Describe the feature and its real job in one sentence.
- List the personal data it might collect.
- Cross out everything the job does not truly need.
- Write the minimized version and one clear consent question for it.
Success looks like
- The feature's job is stated clearly.
- At least one piece of unnecessary personal data is removed.
- A clear consent question is written with an easy way to decline.
Reflection
Saved on this device as you type
Please don't include private information. Your answers are saved only in this browser and are never sent anywhere.
Recap
Privacy means controlling personal data; consent must be informed, and data minimization keeps only what a feature truly needs.
- Personal data is information about a particular person, like a name, address, or exact location.
- Real consent is informed and freely given, not hidden in fine print.
- Data minimization means collecting only what the feature needs to work.
- Data you never collect cannot be leaked or misused, so less data means less risk.
Grades 7–8 extension
The trade-off behind 'free' apps
Many apps are free because they collect and use data about their users. That does not make every free app harmful, but it does mean the question 'what does this app get from me?' is worth asking.
Pick a free app or service and describe the possible trade: what does the user get, and what data might the company collect in return? Which pieces of that data would data minimization say the app does not actually need?
Finish this lesson
- Attempt the knowledge check — not done
- Save a reflection (recommended) — not done
Attempt the knowledge check to finish this lesson.