Skip to main content
45 minutes Intermediate Govern

AI Use Policy (Living)

A working AI use policy for a comms team: what is permitted, what needs review, what is off-limits, how AI use is disclosed and how data is handled. Built to be kept current, with an owner, a version and a review date.

Keep in mind The draft cannot decide data protection or legal questions for you: your data protection lead and Legal must review sections 6, 7 and 10 before it is approved.

Version 1.2 · Updated 30 October 2026

What you'll get

The opening of a worked example. Anonymised and shortened; yours will follow your inputs.

Fictional example.

AI Use Policy, Tarnside Council communications team (extract)

Permitted (no approval): first drafts of internal documents; summarising published reports; reformatting approved copy; brainstorming angles.

Needs review: any copy for residents or the media (reviewed by the duty press officer for accuracy, voice and disclosure); anything citing data (checked against the source before use).

Prohibited: entering residents' personal data or unpublished committee papers into any tool not on the approved register; using AI to produce final sign-off; generating quotes attributed to councillors or officers as if they had said them.

See the full example

Next step

The prompt ends with questions for you. Answer them with this, in the same chat.

Here are my answers to your questions: [YOUR ANSWERS]

Revise the draft using them. Remove the [Assumed] and [Confirm] marks my answers resolve, keep the rest, and update the questions and checks.

Human review checklist

Go deeper

Variations

Stress-test an existing policy

Use this when you already have a policy and want to find its weak points.

Here is our current policy: [PASTE THE POLICY, WITH ITS VERSION AND DATE]. Instead of drafting a policy, review this one twice: first as a team member looking for loopholes, then as a reviewer checking it would stand up to Legal, the data protection lead or a regulator. Give a table with the columns Section | Problem | Why it matters | Proposed fix, covering rules that are unclear, common comms tasks it does not cover, references to tools or terms that may be out of date, and rules nobody could check were followed. Put the five most important first. Quote the policy exactly; do not say a rule is out of date because of a tool change unless I have told you about the change.

One-page note for staff

Use this when the team needs something short now, before the full policy is agreed.

Instead of the full policy, write a one-page AI acceptable use note for staff, under 300 words: what you can use AI for, what needs a second pair of eyes, what you must never do, what information must never go into an AI tool, and who to tell if something goes wrong. Use only rules I have given you; anything else is a proposal. Head it "INTERIM NOTE: replaced by the full policy once agreed".

Full example output

Example output

Anonymised and shortened. Real outputs will vary based on your inputs.

Fictional example.

AI Use Policy, Tarnside Council communications team (extract)

Permitted (no approval): first drafts of internal documents; summarising published reports; reformatting approved copy; brainstorming angles.

Needs review: any copy for residents or the media (reviewed by the duty press officer for accuracy, voice and disclosure); anything citing data (checked against the source before use).

Prohibited: entering residents’ personal data or unpublished committee papers into any tool not on the approved register; using AI to produce final sign-off; generating quotes attributed to councillors or officers as if they had said them.

Disclosure: for consultation summaries published to residents, we say where AI helped produce the summary, in line with our principles.

Accountability: the named owner of each piece is accountable for it. “The AI produced it” is never a defence.

Note: shortened extract. Your policy will reflect your own tools, sector and obligations.

The template

Teams with no AI policy yet can start with the one-page AI acceptable use note and come back to this template when they need the full policy.

AI Use Policy

Organisation: [Name] Applies to: [Team or function] Policy owner: [Name and role] Version: [e.g. 1.0] Approved: [Date] Next review: [Date]

1. Purpose and scope

[One short paragraph: what this policy covers, who it applies to, and that it operationalises the team’s AI Use Principles. Note any tools or activities explicitly out of scope.]

2. Permitted uses (no approval needed)

Tasks the team can use approved AI tools for as a matter of course, within the rules below.

  • [e.g. First drafts of internal documents and briefings]
  • [e.g. Research synthesis and summarising public material]
  • [e.g. Reworking and shortening existing approved copy]
  • [Add your own]

3. Uses that need review first

Tasks where AI may be used, but a named person must review before the work is used or published.

Use caseWho reviewsWhat they check
[e.g. Copy for the public, members or service users][Role][Accuracy, voice, disclosure]
[e.g. External announcements][Role][Accuracy, risk, sign-off]
[e.g. Anything quoting data or statistics][Role][Verified against source]

4. Prohibited uses

Hard lines. AI is not used for these, full stop.

  • [e.g. Final sign-off or approval of any external communication]
  • [e.g. Entering confidential or personal data into non-approved tools]
  • [e.g. Generating quotes attributed to real people as if they had said them]
  • [Add your own]

5. Approved tools

[Reference your AI Tool Register as the live list of approved tools, rather than naming tools here where they will go out of date. State that only tools on the register may be used for work, and how someone requests a new tool be assessed.]

6. Data and confidentiality

[The rules on what information may and may not be entered into AI tools, by sensitivity level. Be specific about unpublished or commercially sensitive material, personal data and anything under NDA. Note which approved tools are cleared for which data levels.]

Approval covers a use: a tool on the register is cleared only for the tasks and kinds of information listed against it. A new use is agreed by [the communications lead] with [our data protection lead, legal adviser or, where there is neither, a named trustee or our sector body].

7. Data protection

Give each point an answer or an owner, and check them with your data protection lead.

  • Lawful basis. [The lawful basis for personal data put into an AI tool (for example, legitimate interests or consent), and whether your privacy notice covers it. Consent to publish a story does not automatically cover AI processing.]
  • Special-category data. [Health, ethnicity, religion, sexual orientation and the other special categories need an extra condition on top of a lawful basis. Keep them out of AI tools without your data protection lead’s agreement.]
  • DPIA triggers. [A data protection impact assessment before a new tool or new use of personal data, and for profiling or large-scale processing. AI counts towards the ICO’s high-risk criteria. Name who runs it.]
  • Retention. [How long you keep AI outputs containing personal data; what the vendor keeps (prompts, files, outputs) and for how long.]
  • Processor terms and transfers. [Whether the vendor works only on your instructions, under a data processing agreement, without training on your data; and what covers any processing outside the UK. Record each tool’s answers in the AI Tool Register.]

8. Disclosure

[When and how AI involvement is disclosed, internally and externally, in line with your principles. Give the standard wording or approach the team should use.]

9. Accountability

[State plainly that a named human is accountable for any output AI helps produce, and that “the AI did it” is never a defence. Note where responsibility sits for review, approval and the policy itself.]

10. When something goes wrong

[A short, blame-aware route for raising an AI-related mistake or near-miss, and where it gets logged. Reference the AI Use Review and Risk Log.]

If you think personal data or confidential material has gone where it should not, stop using that tool for the task and tell [the communications lead] straight away, the same hour if you can: some incidents have to be reported to the ICO without undue delay and, where feasible, within 72 hours of us becoming aware of them, with reasons given for any delay. [The communications lead] brings in [our data protection lead], who assesses whether it must be reported.

11. Review

[State the review cadence (e.g. quarterly), who owns it, and that the version and date at the top are updated on each review.]

When to use it

What it is

The AI Use Policy is the operating layer of your AI governance. Where your AI Use Principles set the stance, this document sets the rules: what the team can use AI for without asking, what needs a second pair of eyes, what is off-limits, how AI involvement is disclosed, and what can and cannot be put into a tool.

The word “living” matters. The risk is a policy written once, signed off, and left to rot while the tools and the team move on. A policy that does not reflect the tools people actually use can end up quietly ignored. This template is built to be maintained: it carries a version, an owner and a review date, and it is designed to be revised on a cadence.

This is the document Legal, IT and leadership want to see, and the one the team can point to when a judgement call is questioned. Keep it short, specific and usable, so people actually open it.

When to use it

Use this template when:

  • You have agreed your AI principles and need to turn them into operating rules
  • Your team keeps asking “is it OK to use AI for this?” and getting inconsistent answers
  • Legal, IT or leadership has asked for a defensible position on AI use
  • An existing policy is out of date and no longer matches the tools in use

Don’t use this template when:

  • You have not agreed your underlying stance (write your AI Use Principles first)
  • You only need to record which tools are in use (use the AI Tool Register)

Inputs needed

  • Your AI Use Principles, so the policy is consistent with the stance you have set
  • An honest list of the AI tools your team uses, sanctioned or not
  • What the team uses AI for now
  • The information you handle and your data protection, legal and regulatory obligations, so the rules are defensible
  • A decision on who owns approvals, who reviews AI-assisted work and who owns the policy itself
  • Who your data protection lead and Legal contact are
  • Anything you have already decided about disclosure

Related templates

AI Tool Register

A living register of the AI tools your comms team uses, and the skills and plugins installed in them: what each is approved for, what data it may handle, who owns it and when it was last reviewed. The single source of truth your AI use policy points to.

Govern

40 minutes · Intermediate

Approval Workflow Mapper

Map out sign-off processes for content approval workflows to ensure clear accountability and timelines.

Govern

18 minutes · Straightforward

AI Use Review & Risk Log

A simple cadence and running log for keeping a comms team's AI use governed over time: a quarterly review of policy, tools and standards, plus a standing record of issues, near-misses and decisions.

Govern

30 minutes per review · Intermediate

AI Use Principles

A short, foundational statement of how your organisation uses AI in communications: the principles that sit above the operating policy and inform every decision about where AI is and is not trusted.

Transform

30 minutes · Straightforward

Applied

Tested in practice

From Applied Comms With AI

All Applied pieces