AI Use Principles

What it is

AI Use Principles is a short foundational document, usually a single page, that sets out what your organisation believes about using AI in communications. It is the stance, not the rulebook: the small number of principles that everything else ladders up to.

This sits deliberately above the detailed AI Use Policy. The policy tells people what they can and cannot do day to day; the principles explain why. When a new tool appears, a new task comes up, or a grey-area judgement has to be made, the policy will not always have an answer, but the principles will give the team something to reason from. Get the principles agreed once, and the working policy, the tool choices and the disclosure decisions all become easier and more consistent.

It is foundational work in the truest sense: you do it once, with the right people in the room, and it serves as the reference point for years. It is also the document that signals, to leadership and to clients, that your AI use is considered and accountable rather than improvised.

When to use it

Use this template when:

• You are setting up AI use in a comms team properly for the first time and want to start from agreed principles
• You already use AI informally and want to put a considered stance underneath it
• You are about to write or commission a detailed AI use policy and need its foundation first
• Leadership, Legal or a client has asked what your position on AI actually is

Don’t use this template when:

• You need the operational detail of what is permitted and prohibited (use the AI Use Policy instead)
• You are assessing whether the organisation is ready to adopt AI at all (use the AI Readiness Assessment first)

Inputs needed

• An honest picture of how AI is being used now, not how you would like it to be used
• Your organisational values or brand voice guidelines, if they exist, so the principles are consistent with them
• Awareness of any client, sector or regulatory obligations that constrain what you can do
• A small group (two or three people) with the authority to agree the principles on the team’s behalf

---

The template

AI Use Principles

Organisation: [Name] Communications function: [Team or department] Agreed by: [Names and roles] Date agreed: [Date] Next review: [Date, suggested every 6 to 12 months]

---

Our stance in one sentence

Write a single sentence that captures how your organisation uses AI in communications. This is the line a team member should be able to recall and apply under pressure.

[For example: “We use AI to do better communications work faster, never to replace human judgement, accountability or the relationships our reputation rests on.”]

---

The principles

Aim for five to eight. Each principle is a short statement plus a sentence of explanation. Adapt the prompts below to your context; delete any that do not fit and add your own.

1. Human accountability [A named person is always accountable for anything AI helps produce. AI assists; people remain responsible.]

2. Transparency and disclosure [State when and how you disclose AI involvement, internally and externally, and the spirit behind that choice.]

3. Accuracy and verification [AI output is a draft to be checked, never a fact to be trusted. State the verification expectation.]

4. Voice and brand integrity [AI must serve your voice, not flatten it. State how you protect tone, character and quality.]

5. Privacy and data care [What information may and may not be put into AI tools, and the principle behind it. Especially important for sensitive or client material.]

6. Fairness and bias awareness [A commitment to checking AI output for bias, stereotyping and exclusion before it goes out.]

7. Proportionate use [Where AI adds genuine value, and where the human-only route is the right one. AI is not the default for everything.]

8. We keep learning [A commitment to reviewing these principles as the tools, risks and our own understanding change.]

---

What these principles rule in and out

A short, plain-English summary to orient anyone reading for the first time.

We are comfortable with	We are cautious about	We do not do
[e.g. AI-assisted first drafts, research synthesis, summarising]	[e.g. client-facing copy, sensitive announcements]	[e.g. final sign-off by AI, entering confidential client data]

---

How these principles are used

[One or two lines on how the principles connect to the rest of your foundations: that the AI Use Policy is built from them, that they are referenced in onboarding, and that team members can raise a principle when something feels off.]

---

AI prompt

Base prompt

I'm drafting a short set of AI Use Principles for a communications team. These principles sit above our detailed policy and capture our stance on using AI in comms work.

About us:
- Organisation and sector: [DESCRIBE]
- Team size and type (in-house / agency): [DESCRIBE]
- How we currently use AI: [DESCRIBE: tools and main tasks]
- Our values or brand voice, if documented: [PASTE OR DESCRIBE]
- Constraints we work within (client, sector, regulatory): [DESCRIBE]

Please draft:
1. A single-sentence statement of our overall stance on AI in communications
2. Six to eight short principles, each a clear statement plus one sentence of explanation, covering at least: human accountability, transparency, accuracy, voice and brand, data and privacy, and bias
3. A short "comfortable with / cautious about / do not do" summary table

Keep the language plain, confident and specific to us. Avoid generic AI-policy boilerplate. These should sound like principles a real team would recognise and use.

Prompt variations

Variation 1: Pressure-test draft principles

Here are our draft AI Use Principles:

[PASTE PRINCIPLES]

Act as a sceptical comms director and an in-house lawyer in turn. For each principle:
1. Is it specific enough to actually guide a decision, or is it a platitude?
2. Where could someone reasonably misinterpret it?
3. What real situation would this principle struggle to resolve?

Then suggest sharper wording for the three weakest principles.

Variation 2: Tailor to a sector

Adapt these draft AI Use Principles for a [SECTOR, e.g. charity / regulated financial services / public sector] communications team:

[PASTE PRINCIPLES]

Flag any principle that should be stronger, more cautious, or worded differently given this sector's obligations and public expectations.

Tips for better AI output:

• Give the model your real values and constraints; generic input produces generic principles
• Ask it to be specific to you and to avoid boilerplate, explicitly
• Always have a human shape the final wording; principles your team did not author are principles your team will not own

---

Human review checklist

• Agreed by the right people: the principles are owned by people with the authority to set them, not drafted in isolation
• Specific, not generic: each principle could only be yours, and would actually guide a real decision
• Consistent with your values: nothing here contradicts your brand voice or organisational values
• Honest about current use: the principles reflect how AI is really used, with a credible path to where you want to be
• Plain English: a new starter could read and apply these without a glossary
• Constraints reflected: client, sector and regulatory obligations are visible in the principles
• Memorable: the one-sentence stance is reliably recallable under pressure
• Reviewable: a review date is set, and the principles acknowledge they will change

Example output

AI Use Principles, Northbank Communications (illustrative)

Our stance: We use AI to raise the quality and speed of our work, never to replace the human judgement and accountability our clients rely on.

1. Human accountability. A named person owns everything we publish. AI never has the last word.
2. Transparency. We are open with clients about where AI supports our work, and we never pass AI output off as bespoke human craft where that matters.
3. Accuracy first. Every AI-generated fact, figure or quote is verified against a source before it leaves the building.
4. Our voice, protected. AI drafts to our voice guidelines; it does not get to set the tone.
5. Careful with data. No confidential client information goes into a tool that is not approved for it.
6. Bias checked. We read AI output for stereotyping and exclusion as standard.
7. Proportionate. Some work is faster and better with AI; some is human-only. We choose deliberately.

Note: illustrative. Your principles will reflect your own values, sector and way of working.

Related templates

• Comms Foundations Index - Record this document as one of your canonical foundations
• AI Readiness Assessment - Establish readiness before setting principles, if you have not already
• AI Use Policy - Turn these principles into the operating rules your team follows day to day
• Brand Voice Audit - Make sure your principles protect a voice you have actually defined

---

Need this set properly for your team? Deploy Comms With AI builds tailored AI principles, policy and governance for comms teams, and Manage Comms With AI keeps them current.