Unions 21 AI principles
These principles guide how Unions 21 uses AI in its work. They are designed to be both a practical framework for internal decision-making and a public statement of our approach.
We hope it will be useful to unions developing their own policies.
1. Put human relationships first, always
Our value to supporter unions is built on relationships. We will never use AI in ways that diminish, hollow out, or replace those, even where doing so might be more efficient. AI can support us in preparing for conversations, but it cannot substitute for being in the room.
2. Always exercise and articulate judgment in using AI
There is no single right answer about when to use AI and when not to. What matters is that we think carefully about each use and can explain our reasoning. If someone asks why we used AI for something, we should always have a good answer. Judgment - not rules - is the through-line of responsible AI use.
3. Use AI to support thinking, never skip it
Insight is a workstream for us and AI should enhance this. We use AI to help us think, for example, structuring ideas, testing arguments, analysing data we understand, but we never delegate the thinking itself. Every piece of work that carries the Unions 21 name reflects genuine human thought and judgment.
4. Be honest about AI use
When we use AI in our work, we are open about it. We don't actively hide AI involvement or pass off AI-generated content as entirely human. At the same time, transparency is a judgment call – not every document needs a disclaimer, but we should never deceive anyone about how something was produced.
5. Recognise AI's impact on people and workers
AI is a politically charged and ethically complex technology, particularly in the union movement where members' livelihoods are directly affected by these changes. We approach AI adoption with care and humility. We are neither dismissive nor zealous. When we talk about AI with unions, we do so with empathy for the real concerns their members have.
6. Protect what's shared in confidence
Unions trust us with sensitive information. We will not share personal data, member union data, or confidential information with AI tools without being certain of the security and legality of doing so.
If it's someone else's information, we treat it with care, regardless of how low the risk may seem.
7. Verify everything
AI is confident but not reliable. We only use AI to work with subjects we know well enough to interrogate the output. We never publish AI-generated content without meaningful human review, and we take responsibility for everything we put our name to, whether AI was involved or not.
8. Learn by doing, together
We are building our understanding of AI through deliberate, shared experimentation. We try things, reflect on what works, and develop our collective confidence. This means everyone in the team understands the tools we use and the decisions we make – no one is left behind and no process is automated without collective agreement.
9. Choose our tools thoughtfully
The AI tools and platforms we use reflect our values. We aspire to move incrementally toward technology choices that are ethical, values-aligned, and consistent with our positions on data sovereignty and workers' rights. This ongoing journey is a conscious direction of travel.
10. Practise what we preach
As an organisation that supports unions in navigating AI, we hold ourselves to the standards we advocate. Our own AI use should be a model for the thoughtful, principled approach we encourage in others.
These principles were developed collaboratively by the Unions 21 team and will be reviewed as our experience with AI evolves. They are intended as a living document.
AI transparency statement: This document was drafted with AI assistance (Claude, Anthropic) based on collaborative discussion and a facilitated team session. The principles reflect decisions made by the Unions 21 team.