Using AI at work: what to check with your employer first
Use AI at work confidently and within the rules, protecting yourself, your employer and your customers.
Before using AI at work, check three things:
- Is there an AI policy? Many employers now have one. Ask your manager or look on the intranet.
- Is there an approved tool? Many organisations provide one, like Microsoft Copilot or a company version of ChatGPT, Claude or Gemini. These usually protect company data better than personal accounts.
- What can't you share? Customer details, colleagues' personal information and confidential documents usually shouldn't go into personal AI accounts.
Safe ways to start:
- drafting emails and documents (without confidential details)
- summarising your own notes
- brainstorming ideas
- explaining jargon or concepts
Always:
- check AI work before you use it; you're responsible for it
- be open about using AI if your workplace expects it
Using AI well is fast becoming a valued work skill. Doing it within the rules protects you.
Why it matters
Using AI at work can save hours a week. But using the wrong tool in the wrong way can cause real problems: leaking confidential information, breaching data protection law, or sending out errors under your name.
The three checks
1. Policy. Many employers now have an AI policy covering which tools are allowed, what they can be used for and what must never be shared. If you can't find one, ask. Asking shows good judgement.
2. Approved tools. Workplace versions of AI assistants (such as Microsoft 365 Copilot, ChatGPT Enterprise or Team, Claude for Work, Gemini for Google Workspace) usually come with contracts that stop your data being used for training and keep it within the organisation's controls. Personal free accounts usually don't.
3. Data rules. As a rule of thumb, keep these out of personal AI accounts:
- customer or client information
- colleagues' personal details (including HR matters)
- financial results before they're public
- contracts, legal advice and trade secrets
- anything marked confidential
Good uses at work
| Task | Tip |
|---|---|
| Emails and documents | Draft with AI, edit in your voice, check facts |
| Meeting notes | Use your approved tool; tell attendees if recording |
| Research and learning | Verify sources before sharing |
| Brainstorming | Great for options; you decide |
| Spreadsheets | Ask it to explain formulas; check results |
| Writing job adverts or reviews | Check for biased wording |
Your responsibilities
- You own the output. If AI gets it wrong and you send it, it's your mistake.
- Transparency. Some workplaces want AI use disclosed, for example in reports or client work.
- Fairness. Be extra careful using AI in decisions about people, such as hiring or performance.
If there's no policy
Suggest one. Meanwhile, use AI only for non-confidential tasks, and keep personal and work use separate.
The workplace landscape
Surveys in the UK and elsewhere consistently find that a large share of workers use AI tools, and that many do so without their employer's knowledge or approval, sometimes called . Employers are responding with policies, approved tools and training. Using AI well, safely and openly is increasingly seen as a core skill rather than a niche one.
Legal considerations in the UK
Data protection (UK GDPR and the Data Protection Act 2018). Entering personal data about customers or colleagues into an AI tool is "processing" that data. Your employer must have a lawful basis, appropriate contracts with the provider, and suitable safeguards. A personal AI account usually provides none of these. The Information Commissioner's Office (ICO) publishes guidance on AI and data protection.
Confidentiality. Employment contracts and client agreements typically restrict sharing confidential information with third parties. Pasting it into a consumer AI tool can breach those terms.
Intellectual property. Material you create at work usually belongs to your employer. Some client contracts restrict the use of AI tools on their work.
Equality law. Using AI in recruitment, promotion or performance decisions risks indirect discrimination if the system is biased. The Equality Act 2010 still applies when a computer is involved, and people may have rights relating to automated decision-making.
Regulated sectors. Finance, healthcare, law and the public sector have extra rules on record-keeping, accountability and data handling.
What good workplace AI looks like
- Enterprise agreements: no training on company data, defined retention, security certifications, data location options.
- Access controls: AI only sees what each employee is allowed to see.
- Clear guidance: which tasks are encouraged, which need review, which are off-limits.
- Training: on prompting, checking outputs and data handling.
- Disclosure norms: when to label AI-assisted work.
Making the case if your workplace lags behind
If your organisation has no approved tool, a practical proposal helps: list three or four specific tasks AI could speed up, estimate the time saved, suggest an enterprise tool with data protections, and propose a small pilot with clear rules. Many organisations adopt AI after one team demonstrates value safely.
Career angle
Employers increasingly value people who can use AI to work faster without compromising quality or confidentiality. Skills worth developing: clear prompting, critical checking, knowing where AI fits in your role, and explaining your AI use to others. Our news tracks how AI is changing work and jobs.
Try this prompt
Help me write a short, friendly message to my manager asking whether we have an approved AI tool and any guidance on using AI for [my main tasks].
Replace the [brackets] with your own details
On haystak
Quick check · 1 of 3
Which is usually fine to put into a personal AI account?
Next step
How AI models work, without the maths
Understand what's really happening when an AI answers you, explained without any maths or code.