Using AI at work policy

AI tools can save time and help you do better work. They can also get things wrong, expose information and treat people unfairly. This policy explains how to use them safely.

AI tools include chat assistants, tools that write text, code or images, and AI features in software we already use. This policy applies whenever you use AI for your work, on any device.

Use approved tools only

Use only the AI tools we have approved for work. IT keeps the list, and can tell you what each tool may be used for.
Sign in with your work account, not a personal one, so that our privacy and security settings apply.
To suggest a new tool, ask IT. Do not use it for work until it has been approved.

Good and bad uses

Examples
Use AI to	Do not use AI to
Draft or improve emails, documents and slides	Send anything to a customer that nobody has read
Summarise long documents you are allowed to read	Summarise documents with personal data in a tool not approved for it
Brainstorm ideas, plans and outlines	Make decisions about people, such as hiring, pay or performance
Explain code, or help you write and test it	Give customers legal, medical or financial advice

Keep information safe

Do not put personal data into an AI tool unless the tool is approved for that use. Personal data is information about a person who can be identified, such as a name, an email address, a photo or an HR record.
Do not put in confidential company information, or information that belongs to our customers or partners, unless the tool is approved for it.
Do not put in passwords, access keys or other security details.
If you are not sure whether information is safe to use, ask your manager or IT first.

Check everything AI produces

AI tools can state wrong things as facts, make up sources and repeat bias. Check every output before you use it. You are responsible for work you produce with AI, just as if you had done it all yourself.

Check facts, figures, quotes and references against reliable sources.
Read and test any code before you use it.
Do not use output that copies someone else's work without permission.

People decide about people

AI must not make decisions about people. A person makes every decision that affects someone's job, pay, performance, discipline or access to our services. You may use AI to help, for example to summarise notes, but the person deciding checks the facts, considers fairness and can explain the decision.

Do not use AI to monitor, profile or score colleagues or customers unless we have assessed and approved that use.

AI can repeat bias against groups of people. When AI helps with anything about people, check that the result is fair to everyone, including disabled people and people whose first language is not English.

Meetings, images and voices

Before you use an AI note-taker or transcription tool in a meeting, tell everyone in the meeting and stop if anyone objects.
Do not use AI note-takers in meetings about someone's health, a grievance or discipline.
Do not use AI to make images, video or audio of real people, including colleagues, without their permission, or to make anything that could mislead people about what is real.

Be open about using AI

Tell people you used AI when it matters to them, for example in work you deliver to a customer, or when they ask.
If a customer or partner has rules about using AI on their work, follow them.
Do not suggest that a person wrote or checked something when they did not.

Connecting AI to our systems

Do not connect an AI tool to company email, files, calendars or other systems unless IT has approved the connection.
Do not let an AI tool act for you, such as sending messages, making payments or changing records, unless we have approved that use.
Content from outside the company, such as emails, documents and web pages, can hide instructions that trick an AI tool. This is called prompt injection. Report anything suspicious to IT.

If something goes wrong

Tell your manager or IT straight away if something goes wrong: for example, if you put information into an AI tool that should not be there, or someone relied on a wrong answer. If personal data was involved, it may be a personal data breach, so report it at once, as our Data protection at work policy explains.

Report honest mistakes quickly. It is much easier to put things right early.

Learning to use AI well

We provide training on using AI tools safely. Complete it before you use AI for work, and again when we ask you to. Ask IT or your manager if you need more help.

If this policy is broken

Breaking this policy may lead to disciplinary action under our disciplinary procedure. Serious or repeated breaches may lead to dismissal.

Find out more

Confirming you have read this policy

AI tools change quickly, so we review this policy and the approved list at least every 6 months, and tell you what has changed. Ask IT or your manager if anything in it is unclear, then confirm below that you have read it.