Most companies have values on the wall. Few have translated them into practical rules for AI. Yet AI forces value choices every week: how much to tell customers, what to automate, how to treat staff whose work changes, which data to use. The way to make AI decisions consistent with your values is to turn each value into specific AI practices, agree a few red lines, and use a short set of questions whenever a new AI use is proposed.
Step 1: Map values to AI practices
Take each of your stated values and ask, "what would this mean for how we use AI?" For example:
| Company value | What it means for AI |
|---|---|
| Honesty | We tell customers when they are interacting with AI or receiving AI-generated content where it matters to them |
| Respect for people | We involve staff in redesigning their work and are open about how AI affects roles |
| Quality | A qualified person reviews AI output before it reaches customers or decisions |
| Client trust | Client information is only used in approved tools with suitable privacy terms |
| Fairness | We do not let AI make or heavily influence decisions about people without human accountability |
This exercise often reveals that your values already give clear guidance. They just need to be written down for AI.
Step 2: Agree your red lines
Red lines are uses you will not allow, whatever the business case. Common examples:
- No fully automated decisions about hiring, firing or discipline.
- No confidential client data in unapproved tools.
- No AI-generated content presented as personal human advice where that would mislead.
- No monitoring of staff through AI without transparency and a lawful basis.
Keep the list short and clear. A few firm lines are more useful than many vague principles.
Step 3: Use five review questions
When someone proposes a new AI use, ask:
- Who could be harmed, and how?
- Would we be comfortable if customers or staff knew exactly how we use AI here?
- Who is accountable for the outcome?
- Could this treat some people unfairly?
- Does it comply with UK GDPR and any sector rules?
Most proposals pass quickly. The few that do not are exactly the ones that needed scrutiny.
Making it practical, not bureaucratic
- Put the values map and red lines in your AI use guide, written in plain English.
- Let low-risk uses proceed without formal review.
- Reserve deeper review for uses involving customers, personal data or decisions about people.
- Revisit the map annually as AI capabilities change.
Ethics is a leadership responsibility
AI ethics is sometimes treated as a specialist or legal topic. In practice, most ethical AI decisions are everyday leadership choices made by managers. That is why managers need to understand both what AI can do and where its risks lie. Our guide to AI ethics every leader should know covers the principles in more depth, and AI ethics for business looks at common dilemmas.
Our Leading AI Innovation & Organisational Transformation course includes using AI safely and responsibly, checking for errors and bias, and drafting simple AI guidelines for your team.
Frequently asked questions
Do we need a separate AI ethics policy?
Not necessarily. Many organisations fold AI principles into existing policies and an AI use guide. What matters is that the rules are clear and known.
Should we tell customers when we use AI?
Where it affects them meaningfully, transparency is usually the right choice and fits most company values. Some sectors have specific disclosure expectations.
Who should decide AI red lines?
The leadership team, with input from staff, legal and data protection. Red lines should reflect what the organisation stands for.