THE FUTURE MARKETACADEMY
Your classroomRoom 01 · AI Lab
School of AI

AI Risk at Work: privacy, bias and misuse in practice

Step 5 of 7 in AI Lab

Next in your path: 1.2.2 Prompting for Complex Tasks: chains, roles and structured outputs

What you'll be able to do
  • Sort workplace AI risks into three families: privacy, bias and security or misuse.
  • Explain, in plain words, what privacy regulators in your country expect when a team uses AI with personal information (with Canadian and Philippine examples).
  • Know what a privacy breach is, when it must be reported, and what to do in the first hour.
  • Spot unfair results when AI helps with decisions about people, and keep a human in charge.
  • Recognize prompt injection, AI-powered phishing and other misuse, and shut the easy doors.
  • Complete a simple one-page AI risk check and write clear do and don't rules for a team.

Before this course: AI Safety, Privacy and Ethics for Everyday Users (course 1.8). You already know how to change privacy settings, what never goes into a prompt, how to check an answer, how to spot AI scams, and how to turn on MFA. This course moves from your own habits to the workplace: real jobs, other people's data, and rules a team can follow. Education only; this is not legal advice. For a real incident or a legal question, talk to your privacy officer or a lawyer.

Words in this course 49
AgentAI (artificial intelligence)APIArticleCustomer acquisition cost (CAC)Data analyticsData breachData minimisationDefinition of doneDemerit pointsDesign briefFeedbackFillForceGoogle FormHTMLHTTPSLarge language model (LLM)LetMulti-factor authenticationMulti-factor authentication (MFA)Necessity and proportionalityPDFPhishingPhishing (cybersecurity)PIPEDAPlain formPrivacy settingsProjectProject (Claude)PromptPrompt chainPrompt injectionRestRetention (audience retention)RiskRisk responseShow, tell, trySocial engineeringSpread
Open the Vocabulary Hall
01

The three families of AI risk at work

In simple words

Think of AI at work like a new kitchen helper. They might leak the secret family recipe (privacy), they might treat some customers unfairly without noticing (bias), or someone might trick them into opening the back door (security and misuse). Knowing the three kinds of trouble helps you look in the right places.

Key facts
  • The Canadian Centre for Cyber Security (guidance ITSAP.00.041) names these generative AI risks, among others:
  • Privacy of data: users may unknowingly enter sensitive company data or personal information into prompts.
  • Biased content: training data from the open internet and unbalanced representation can make output prejudiced.
  • Misinformation and disinformation: AI content that isn't clearly labelled can spread false information.
  • Phishing: attackers can write more frequent, automated and convincing spear-phishing messages.
  • Malicious code and buggy code: attackers can create malware, and developers can introduce insecure code.
  • Poisoned datasets: malicious content in training data can weaken output quality and raise supply-chain risk.
  • Loss of intellectual property: AI can help attackers steal company data more easily.
  • This course groups them in three families:
FamilyThe question it asksExample at work
PrivacyWhose information is going in or coming out?A support agent pastes a customer's full file into a chat
BiasCould the result treat some people unfairly?AI ranks job applicants and favours one group
Security and misuseCould someone trick or abuse the tool?A web page hides instructions for an AI agent
  • The Cyber Centre's advice to organizations includes setting policies on how AI is used, choosing vendors carefully, being careful about what information goes in, using MFA, applying updates, and training employees.
Step by step
  1. Write down every way your team already uses AI (writing, summaries, search, spreadsheets, customer replies). Include tools people use on their own.
  2. For each use, write the kind of information that goes in: public, internal, or personal.
  3. Label each use with one family that worries you most: Privacy, Bias, or Security and misuse.
  4. Circle any use that touches personal information or decisions about people.
  5. Put the circled uses at the top. These are the ones you will examine first in the next lessons.
  6. Check: every use has an information type and a family label, and at least one is circled or you can explain why none is.
Watch on YouTube
AI privacy: what happens to your data when you use AI tools?CampusGuard · YouTube · youtube.com

Videos by other creators on YouTube, for reference. They are not made by or affiliated with The Future Market.

LESSON 01The three families of AI risk at work
LESSON 01 / ANIMATED LESSON / AI VOICE00:00 / 01:00
Read the video script
[Host] "AI at work can go wrong in three ways. Learn them and you know where to look."
[Text on screen: 1. Privacy] "One: privacy. Someone pastes in information that should not leave the building."
[Text on screen: 2. Bias] "Two: bias. The result treats some people unfairly, and nobody notices."
[Text on screen: 3. Security and misuse] "Three: security. Someone tricks the tool, or uses it to trick you."
[Host] "Name the family first. Then you can fix the right thing."
Try it now

Make a one-page list of your team's AI uses with the information type and family for each, and circle the top three to review.

ClaudeStuck? Copy a ready-made prompt for Claude

Copy this, open Claude, paste it, and Claude will walk you through the lesson step by step.

For a grown-up: Claude is for adults (18+). Copy this, paste it into Claude, and use the answer to explain the lesson to your child.

I'm learning "AI Risk at Work: privacy, bias and misuse in practice" on The Future Market. Lesson 1: "The three families of AI risk at work".

What the lesson teaches:
- The Canadian Centre for Cyber Security (guidance ITSAP.00.041) names these generative AI risks, among others:
- Privacy of data: users may unknowingly enter sensitive company data or personal information into prompts.
- Biased content: training data from the open internet and unbalanced representation can make output prejudiced.
- Misinformation and disinformation: AI content that isn't clearly labelled can spread false information.
- Phishing: attackers can write more frequent, automated and convincing spear-phishing messages.
- Malicious code and buggy code: attackers can create malware, and developers can introduce insecure code.
- Poisoned datasets: malicious content in training data can weaken output quality and raise supply-chain risk.

The exercise I'm working on:
Make a one-page list of your team's AI uses with the information type and family for each, and circle the top three to review.

I'm stuck. Please:
1. Explain the main idea in plain, simple words with one everyday example.
2. Walk me through the exercise one small step at a time. Ask me what I've done so far before giving the next step.
3. Don't just give me the final answer. Help me do it myself, and check my work when I share it.
I'm a parent helping my child (about 8 years old) with a lesson called "The three families of AI risk at work" from the course "AI Risk at Work: privacy, bias and misuse in practice".

Here's what the lesson says:
Think of AI at work like a new kitchen helper. They might leak the secret family recipe (privacy), they might treat some customers unfairly without noticing (bias), or someone might trick them into opening the back door (security and misuse). Knowing the three kinds of trouble helps you look in the right places.

The activity:
Make a one-page list of your team's AI uses with the information type and family for each, and circle the top three to review.

Please explain this so an 8-year-old understands: short sentences, a fun everyday example, and no hard words.
Then give me 3 simple questions I can ask my child to check they understood, and a way to do the activity together, one tiny step at a time.
Open Claude
Sources
My notes
02

Privacy at work: what regulators expect

This lesson is part of full access: every course and every level for one payment.

03

When something leaks: privacy breaches

This lesson is part of full access: every course and every level for one payment.

04

Bias in practice: decisions about people

This lesson is part of full access: every course and every level for one payment.

05

Security and misuse: tricks, scams and bad code

This lesson is part of full access: every course and every level for one payment.

06

A simple AI risk check

This lesson is part of full access: every course and every level for one payment.

07

Do and don't rules for your team

This lesson is part of full access: every course and every level for one payment.

++

Final project

This lesson is part of full access: every course and every level for one payment.

??

Quiz

  1. Name the three risk families used in this course.

  2. According to good-practice AI guidelines, when should personal information go into a prompt?

  3. Under Anthropic's commercial terms, who is the data controller and who is the processor?

  4. When should an organization report a breach to the regulator and notify people?

  5. What two things does a real risk of significant harm depend on?

  6. What should you do with records of data breaches?

  7. Many countries set short deadlines to notify the regulator of a breach. What should you do?

  8. Who must make the final decision when AI helps with hiring or grading?

  9. What is prompt injection?

  10. What is a safe starting rule for an AI tool that can take actions?

EX

Final exam

This lesson is part of full access: every course and every level for one payment.

CE

Certificate

This lesson is part of full access: every course and every level for one payment.