Module 4 — Safe AI Use at Work
Your staff are already using AI. The question is what they are pasting into it.
Banning AI does not work — it just moves the usage somewhere you cannot see. This module sets clear, practical boundaries people can actually follow.
- Know precisely what may and may not be put into an AI tool
- Tell the difference between a consumer and a business AI account
- Recognise a hallucinated answer, including fabricated sources
- Verify AI output before it reaches a client or a regulator
- 7 lessonsIncluded
- Paste-or-not decision drillIncluded
- Hallucination detectionIncluded
- Linked campaignSafe AI use
- Acknowledgement recordIncluded
What is in the module
Lesson by lesson
- 1
What actually happens to what you paste
The difference between a consumer account, a business account and an enterprise tenant — and why that difference decides whether your data trains a model.
- 2
What never goes into an AI tool
Client identifiable information, credentials, unreleased financials, health information, legal advice, source code with secrets in it, and anything under NDA.
- 3
Client confidentiality
For professional services, this is the one that matters. Pasting a client’s document into a consumer AI tool can be a disclosure, regardless of intent.
- 4
Business data exposure without meaning to
The subtle version: summarising a spreadsheet, rewriting a contract, or asking for help with an email that quotes a customer complaint.
- 5
Hallucinations, and why confidence means nothing
AI tools produce fluent, well-formatted, entirely fabricated answers — including citations, case law, product features and prices.
- 6
How to verify AI output
A practical checklist. What must be verified before it leaves the building, and what genuinely does not need checking.
- 7
Approved tools and responsible use
Why an approved tool list beats a ban, how to request a new tool, and what to do if you have already pasted something you should not have.
Module details
- Audience
- All staff, and managers setting policy
- Duration
- About 20 minutes
- Pass mark
- 80% to pass
- On completion
- Verifiable certificate
- Format
- Lessons, scenarios, quiz
The full module, its scenarios, the quiz and the certificate live inside the platform. Our blog covers these topics at a high level for free — but the training itself is not published publicly.
See plans that include itWhat people can do afterwards
Outcomes, not attendance
- Know precisely what may and may not be put into an AI tool
- Tell the difference between a consumer and a business AI account
- Recognise a hallucinated answer, including fabricated sources
- Verify AI output before it reaches a client or a regulator
- Use AI for the tasks it is genuinely good at
- Request approval for a new AI tool through the right channel
- Report an accidental disclosure quickly instead of hiding it
- Understand why "I only used it for a summary" is not a safe default
A scenario from the module
Just helping with a summary
A team member has a forty-page contract from a client and a deadline. They paste the whole document into a free AI tool and ask for a summary of the key obligations. The summary is excellent, accurate and saves two hours. Nobody outside the business ever sees the summary, and the team member deletes the chat afterwards.
Nothing visibly went wrong. Name the two things that did — and explain which one would matter most if the client asked about it.
The answer, the explanation and the rest of the scenarios are inside the module.
Who it is for
Assign it to the right people
Professional services firms
Accounting, legal, financial advice and consulting — where client confidentiality is a professional obligation, not a preference.
Any business writing an AI policy
Pair the module with an acknowledgement so you can evidence that staff read and accepted the boundaries.
Managers approving AI tools
Understand what you are agreeing to when you approve a tool for the team to use.
Free vs paid
What you can read for free, and what needs an account.
Free on the blog, no account needed
Understand the issue. Genuinely useful, genuinely free.
- Safe AI use at work: what managers need to understand
- High-level commentary on AI risk in business
- General guidance on verifying AI output
- Why an approved tool list beats a blanket ban
Inside the platform, with a paid account
Act on the issue, and have a record that you did.
- The full seven-lesson Safe AI Use module
- The quiz, pass mark and certificate
- A Safe AI Use campaign with a staff acknowledgement
- The safe AI use risk explainer and recommendation template
- Acknowledgement records showing who accepted the boundaries and when
Training is one stage of five
Where this module sits in the wider platform.
A module on its own teaches. It becomes a risk control when it is assigned off the back of an assessment gap or a failed simulation, and when the completion ends up in an evidence record.
Assessments
Find the gaps that tell you which training to assign in the first place.
Phishing simulations
Test whether it stuck, and assign follow-up training automatically.
How the training works
Scenario decisions, measured competencies, badges and the full catalogue.
Reporting and evidence
Verifiable certificates and completion records for insurers and clients.
Set the boundary before someone finds it the hard way.
See the module and the Safe AI Use campaign, including the staff acknowledgement record.