It’s 4:40 on a Thursday.

You have a report to summarize, an email to soften, and a meeting at 9 a.m. you haven’t prepped for.

The chatbot is right there in another tab. It’s fast, it’s polite, and it never sighs at you.

So you paste in the whole report. Then the client’s email thread. Then, why not, the spreadsheet with everyone’s names in it.

And somewhere around the third paste, a quieter thought arrives: am I allowed to do this?

Fair question. Learning how to use AI at work isn’t really about prompts. It’s about knowing where the lines are, so the tool saves you time instead of creating a problem you have to explain later.

Here’s the plain-English version: what it’s genuinely good for, what never goes in, and three rules that cover most of the risk.

(New to AI generally? Start with What AI Is Actually Useful for in Real Life Right Now.)

Step one: find out if your workplace has an AI policy

Before you type a single prompt, find the rules.

Alberta’s government career site, ALIS, puts it simply: your employer “has likely decided or will decide on rules for using AI in your workplace,” and understanding those rules is the first step.

Where to look:

• Your employee handbook or intranet. Search for “AI,” “generative AI,” “acceptable use” or “ChatGPT.”
• Your IT or privacy team. Ask which tools are approved, and whether there’s a company version.
• Your manager. Especially if you’re in a client-facing, regulated or confidential role.

Questions worth asking:

1. Which AI tools are approved for work, and which aren’t?
2. Is there an enterprise or company account I should use instead of a personal one?
3. What kinds of information are off limits?
4. Do I need to tell anyone when I’ve used AI?

If the answer is “we don’t have a policy yet,” that isn’t a green light. It’s a reason to be more careful, and to stick to the low-risk uses below.

Microsoft’s own workplace safety advice says the same thing in five words: “Use only company-authorized AI services.”

Can my employer see what I type into AI?

Possibly. It depends on the tool, the account and your employer.

If you’re using a company account, a work device or a tool your employer set up, assume your activity could be logged or reviewed, the same way work email can be.

In Ontario, employers with 25 or more employees must have a written policy that says whether they electronically monitor employees, and if so, how and why. The province is clear that the rule is about transparency: it does “not establish a right for employees not to be electronically monitored.”

Rules differ across provinces and U.S. states. The practical rule is the same everywhere: don’t type anything into a work tool that you’d be uncomfortable having your employer read.

Pasting work material into your own account can be a bigger problem, not a smaller one, because your employer has no control over where that data goes.

How to use AI at work: what it’s actually good for

Now the good part. Used well, AI is a very fast first-draft machine and a patient thinking partner.

Low-risk, genuinely useful tasks:

• First drafts. A tricky email, a job posting, a project update, a polite “no.”
• Rewording. Make it shorter, warmer, firmer, clearer, less like 11 p.m. you wrote it.
• Summarizing your own notes. Your messy bullet points from a meeting become a tidy recap.
• Meeting prep. “What questions might a skeptical finance director ask about this plan?”
• Brainstorming. Names, angles, agenda ideas, ways to explain something to a new team member.
• Learning something new. Explain a spreadsheet formula, a term in a contract, a concept in your field. (We wrote about why learning later in life is worth it in Learning Something New After 40.)

ALIS has a smart suggestion while you’re learning: use AI “behind the scenes” first, on internal work, rather than for content customers will see.

What never to paste into a chatbot at work

This is where people get into trouble. Not because they meant to, but because pasting is so easy.

This is where people get into trouble. Not because they meant to, but because pasting is so easy.

Canada’s federal guide for public servants is blunt: “Public servants must not input personal information into publicly available online generative AI tools.” You probably don’t work for the government, but it’s a useful benchmark for everyone.

The Canadian Centre for Cyber Security warns that users “may unknowingly provide sensitive corporate data or personally identifiable information” in their prompts, and advises: “Avoid providing PII or sensitive corporate data as part of the queries or prompts.” (PII means personally identifiable information.)

Keep these out, unless your employer has explicitly approved a tool for them:

• Other people’s personal information. Clients, patients, students, employees, applicants. Names, contact details, health details, performance issues, salaries.
• Confidential documents. Contracts, financials, strategy decks, anything under a non-disclosure agreement.
• HR and medical records. Including your own, if it’s a work tool.
• Passwords, account numbers and access details. Ever. For anything.
• Unreleased plans. Product launches, restructurings, the reorg everyone’s whispering about.

One trap: deleting names isn’t always enough. ALIS points out that even when you remove identifying details, “it may still be possible to link it to an individual.” A “45-year-old store manager in Red Deer who went on leave in March” is not exactly anonymous.

Canada’s privacy commissioners made a similar point in their 2023 principles for generative AI, which encourage people to avoid “entering personal information into a prompt, unless necessary.” (For the personal side of this, our AI privacy guide goes deeper.)

Check everything: AI gets things wrong, confidently

This is the part people underestimate.

The U.S. National Institute of Standards and Technology lists “confabulation” as a key risk of generative AI, defining it as “the production of confidently stated but erroneous or false content.” You’ve probably heard it called “hallucination.”

It isn’t a theoretical problem. Two real Canadian cases:

• The fake court cases. In Zhang v. Chen (2024 BCSC 285), a B.C. lawyer filed an application that relied on two cases that turned out not to exist. ChatGPT had invented them. The court ordered the lawyer to personally pay costs for the extra work it caused, and described citing fake cases as “an abuse of process and is tantamount to making a false statement to the court.”
• The chatbot that made up a policy. In Moffatt v. Air Canada (2024 BCCRT 149), B.C.’s Civil Resolution Tribunal found the airline responsible after its website chatbot gave a customer wrong information about bereavement fares. Air Canada argued it couldn’t be held liable for what the chatbot said. The tribunal disagreed: “It should be obvious to Air Canada that it is responsible for all the information on its website.”

The lesson for the rest of us: if AI helps you do the work, you’re still the one who signs it. ALIS says it plainly: “you need to take responsibility for AI’s work.”

How to check without doubling your workload:

• Verify facts, numbers, names and dates against the original source, every time.
• Ask for sources, then open them. If a source doesn’t exist or doesn’t say what the AI claims, that’s your answer.
• Read it as your manager would. Does it sound like you? Is anything vague, overconfident or slightly off?
• Watch for bias. ALIS warns AI can reflect “biased or discriminatory information” in its output, which matters a lot for anything involving hiring, performance or customers.

Being open about using it

Do you have to tell people you used AI? It depends on your workplace, your field and what you’re producing. Some employers and professions now have disclosure rules; many don’t yet.

The Government of Canada’s guide for its own staff includes “Transparent” in its FASTER principles: “identify content that has been produced using generative AI.” It’s a sensible default for anything public-facing, client-facing or that someone will rely on as your professional judgment.

A good middle ground:

• Internal drafts you rewrite yourself: usually no announcement needed, unless your policy says otherwise.
• Anything published, client-facing or decision-making: follow your policy, and if there isn’t one, ask.
• Never present AI output as research, data or expert analysis you didn’t check.

Will AI take my job?

Nobody can tell you that with certainty, and be wary of anyone who claims to.

What we do know: a 2024 Statistics Canada analysis estimated that many Canadian jobs are highly “exposed” to AI, but noted that “relatively higher occupational exposure to AI does not necessarily imply a higher risk of job loss.” In many roles, AI is more likely to change tasks than replace people.

The practical move is the same either way: get comfortable with the tools your workplace allows, keep your judgment sharp, and keep learning. (If you’re weighing a new role, our guide on how to negotiate a job offer can help.)

Prompts you can copy

Swap the details in brackets. Notice what’s missing: real names, client details and confidential numbers.

• Soften an email: “Rewrite this email so it’s polite but firm, under 120 words. The goal is to push a deadline back one week. [Paste your own draft, with names removed.]”
• Summarize your notes: “Turn these meeting notes into a short recap with decisions, owners and next steps. [Paste your notes, with names replaced by roles.]”
• Prep for a meeting: “I’m presenting a plan to [reduce overtime costs] to a skeptical [finance director]. What are the five toughest questions I might get, and how could I answer each?”
• Explain something: “Explain what [a pivot table] is to someone who uses spreadsheets but isn’t technical, with one simple example.”
• Brainstorm: “Give me 10 agenda ideas for a 45-minute team planning session for a [small customer service team].”
• Check your own work: “Here’s my draft. What’s unclear, what’s missing, and what might a reader push back on?”

Then do the most important step yourself: read it, fix it, and make it yours.

The three rules, on one sticky note

Need a reminder on the desk? Our “Still Learning. Always.” Desk Mug lives in the Muddy Boutique.

The Curious One’s take

AI is a little like a brilliant new intern with no sense of what’s confidential and a habit of making things up when it doesn’t know.

Brilliant interns are wonderful. You still don’t hand them the client files on day one, and you still read their work before it goes out.

Use it for the drafts, the rewording, the 4:40 p.m. brain fog. Keep the judgment, the people and the secrets for yourself.

That’s not being cautious. That’s being good at your job.

More on AI, culture and the digital life we’re all living is on our Growth & Culture desk.

ASK MUDDY

Does your workplace allow AI tools? What do you actually use them for, and what’s off limits?

JOIN THE MUDDY COMMUNITY

Life gets messy. We talk about it. Join the free Muddy community at muddystilettos.ca for new stories every day on the tech, culture and ideas shaping women’s lives.

ABOUT THE MUDDY GROWTH & CULTURE DESK

The Muddy Growth & Culture Desk covers books, shows, AI, culture and the internet rabbit holes worth falling into, and connects them to the bigger questions in women’s lives. We separate reporting from opinion and go back to the original sources.

Featuring The Curious One: books, Netflix, AI, culture and internet rabbit holes. Curious, reflective, sharp.

This article is educational and informational and is not legal or employment advice. Workplace policies, privacy and employment laws vary by employer, province and state and change over time; check your employer’s policy and the rules that apply to you, or consult a qualified professional.

Sources

• How to use AI safely and responsibly at work, Government of Alberta, ALIS
• Safety tips for using AI at work, Microsoft Support
• Your guide to the Employment Standards Act: Written policy on electronic monitoring of employees, Government of Ontario
• Guide on the use of generative artificial intelligence, Government of Canada, Treasury Board of Canada Secretariat
• Generative artificial intelligence (AI), ITSAP.00.041, Canadian Centre for Cyber Security
• Principles for responsible, trustworthy and privacy-protective generative AI technologies, Office of the Privacy Commissioner of Canada
• Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1), U.S. National Institute of Standards and Technology
• Zhang v. Chen, 2024 BCSC 285 (summary), Fasken
• Moffatt v. Air Canada, 2024 BCCRT 149 (summary), Daily Hive
• Experimental estimates of potential artificial intelligence occupational exposure in Canada, Statistics Canada