Responsible AI Use In Practice – Module 4 Security Thinking AI Safety Basics

📍 Module 4: Security Thinking – AI Safety

đź“ť Overview

While many AI tools are easy to use, they still come with security risks that are worth understanding — especially when used in a workplace setting. This module introduces common threats related to AI, safe habits for general users, and a few basic steps that IT or digital leads can take to protect systems and data.

You don’t need to be technical to improve security — most of the steps covered here are simple and practical.

🎯 Learning Objective

By the end of this module, you’ll be able to:

  • Recognise common AI-related security risks

  • Understand how AI tools could be misused — intentionally or accidentally

  • Identify simple habits that help reduce security issues

  • Know when to pause and report concerns

  • Feel more confident using AI tools safely in everyday settings

đź“– Text version of the lesson

Why AI security matters
Like any digital tool, AI systems can be vulnerable to misuse, attack, or human error. This doesn’t mean AI is unsafe — but it does mean we need to use it thoughtfully, especially when sensitive data or business decisions are involved.

Common AI-related threats
Here are some risks that come up in real-world use:

  • Prompt injection – Someone adds hidden instructions into content to change how the AI behaves (e.g. “ignore previous instructions and do X instead”).

  • Data poisoning – Deliberate manipulation of training data to make AI behave incorrectly.

  • Output misuse – Using AI-generated content without checking can lead to reputational or legal risk (e.g. errors, false claims, offensive language).

Risks in everyday use

  • A user pastes sensitive customer info into a free tool — not realising it may be stored or logged.

  • A shared document includes an embedded prompt that manipulates an AI summary tool.

  • An AI-generated image or text is shared without being reviewed — and contains misleading or inappropriate content.

These things are avoidable with a few small changes to how tools are used.

What general users can do

  • Avoid entering sensitive data (e.g. full names, account info) into AI tools, especially free or public versions.

  • Always read and review AI outputs before sending, publishing, or acting on them.

  • Know what tools your organisation is using — and who’s using them.

  • Ask: is this tool storing or reusing data? What does the privacy policy say?

What IT or digital leads can do

  • Monitor which tools are being used across the team or organisation.

  • Set clear rules about tool access, especially in high-risk areas (HR, finance, legal).

  • Enable logging and tracking to help review issues if something goes wrong.

  • Encourage staff to report anything that looks or feels unusual.

If something goes wrong

  • Pause use of the tool

  • Inform your IT or data lead

  • Write down what happened — including what was input and what came out

  • Check whether the issue needs to be reported internally or externally

  • Use the experience to improve future guidance or awareness

Checklist: Quick wins for safer AI use

  • âś… Know what tools are being used

  • âś… Limit who can input or view sensitive data

  • âś… Review outputs before using them

  • âś… Brief teams on good habits

  • âś… Create a simple plan for responding to issues

When to be especially cautious
Extra care may be needed when AI is used in areas like:

  • Hiring and HR

  • Financial planning or credit checks

  • Legal, medical, or educational advice

  • Any system that affects someone’s access to support or services

In these areas, consider getting expert advice — especially if the tool is embedded into your decision-making process.

🪞 Reflective Prompt (Optional)

Think about an AI tool you’ve used or considered using at work.

Question:

What’s the worst-case scenario if that tool didn’t work properly or was misused?

  • What could go wrong?

  • Who might be affected?

  • Do you know who you’d report that to?

Even thinking through these questions can help you spot risks early and respond more confidently.

âś… Suggested Next Step

We recommend reviewing how AI tools are currently used in your team or organisation, and asking:

  • Does everyone know the limits of what these tools can do?

  • Have you briefed your team on what not to input?

  • Is there someone responsible for tracking or managing tool use?

Even a short team discussion or shared document can improve awareness and prevent problems before they happen.

Previous Lesson
Back to Course
Next Lesson