AI can scale part of our campaigning work, but it can’t see the full picture like campaigners do. Here are five things where I, as an AI expert, genuinely think AI can create value for campaigners and four non-negotiable ground rules.
TL;DR: The real test of using AI in campaigning is not how much automation or tasks we hand over to AI to do on our behalf. It is knowing when to bring AI in and when to keep it away.
A year ago, I was convinced that if I did enough coding and created enough agents, I could run the whole campaign strategy process with a press of a button. Now as my understanding of AI matures, I won’t even think about it.
I often hear campaigners debating whether to engage with AI at all. For me, the better question is, “Where can AI boost impact and how to use it ethically?” This allows selectivity and caution, while using the strengths of the technology.
This is my take onsetting up an ethical foundation for using AI for social change.
If I were the Chief AI Officer of a campaign, a job that exists nowhere except in my head, here’s what I’d consider bringing AI and the lines that I’d never cross.
Ground Rules
As a Chief AI Officer, my first executive decision would set up four rules, non-negotiable, before anyone touches an AI tool:
1. If you don’t need AI, don’t use it.
AI for social change doesn’t happen unless AI can help us get what we can’t do alone. Unnecessary use of AI wastes budget and causes unnecessary environmental harm.
2. If a wrong output can put someone at risk, don’t use AI.
The UK’s DWP’s own analysis confirmed that its automated benefit decisions discriminated against disabled people, older claimants, and those from certain nationalities
3. If you can’t verify the output or don’t have time for it, don’t use AI.
Wrong information can damage your credibility, hand your opponents an easy attack line, and may harm the communities you’re trying to serve. Be careful.
4. If you can’t use AI without telling a secret, don’t use it.
When you share sensitive information with AI tools, you risk exposing campaign strategies, donor details, and vulnerable community members’ information to third parties.
There is no point in spending time, money, and environmental cost on AI if it ends up harming the campaign strategy instead of supporting it.
(For more information on ethical use of AI for campaigners, sign up for the Deep Dive on 18 June.)
Where I’d explore with the team using AI for social change:
AI brings two strengths: finding patterns in large amounts of information and using patterns to generate outputs. These are five applications I’d explore with the team:
1. Tailoring content for localisation and accessibility. LLMs are useful for adapting tone, format, reading difficulty and language. One signed off message quickly becomes tailored versions, a complicated policy topic becomes everyday language. A human check will be needed to verify accuracy.
(Top tip: WhatGov, an app that turns parliamentary debates into Gen Z language, is an example of using AI for accessibility. Check it out to get a feeling of what’s possible.)
2. Finding patterns in a large collection of information (reports, consultations, news, action-tracking records, etc). Within seconds, LLMs can do the heavy lift identifying patterns. The campaigner’s role becomes guiding the LLM and interpreting what it surfaces.
(Top tip: “Habermas Machine”, an AI research tool by Google’s Deepmind, shows the synthesis opportunity. It looks for patterns to find a common ground across many perspectives. Such approach can identify messages to test, while campaigners choose which argument to use.)
3. Using publicly-available information to get insights, learning from previous patterns and flagging what is shifting. What people are saying online, what arguments are shifting, where new ideologies are showing. Bias mitigation should be baked into the prompt and what input is being used.
(Top tip: The Drumbeat Machine, comms intelligent tool by Rootcause, shows the synthesis opportunity. It analyses news, online conversations, and influencing insights to produce daily intel tailored to each campaign.)
4. Monitoring and flagging patterns at scale. Think about the need of monitoring thousands and millions of observations. You can teach AI patterns and let it flag when something happens.
(Two examples of this done well: Amnesty International using open-source methods and satellite imagery to document destruction in hard to access war zones. Full Fact, using AI to match repeated misinformation claims. Both examples take advantage of the AI’s ability to verify at scale. AI scales the work, but the verification and judgement stay with the experts.)
5. Visualising what doesn’t exist in our world. There’s still debate on the ethical use of AI to generate content, early results of the 2026 charity digital skills report found 40% of charities using AI to generate social media content. From what I saw, the best uses so far were around visualising the future and demonstrating concepts that can be difficult for audiences to visualise themselves.
(Example: Greenpeace ‘Dare to Dream’ initiative is old but gold in demonstrating how AI can show imagination. Youth from Indonesia used Generative AI to visualise their dreams for the future of the country.)
All five uses demonstrate what AI for social change can look like in practice. They pass the Ground Rules, play to AI’s strengths, scale the work beyond what teams could do alone, and can be checked for accuracy. (For more examples and practice of how social change campaigners can use AI, sign up for the Deep Dive here.)
Over to You
The campaigners who use AI well won’t be the ones that hand over the work to AI, instead the ones who understand the place of AI. This guide isn’t finished. AI is moving fast, and so is our understanding of where it belongs.
Want to learn more about how to use AI effectively in your campaigning, join my next online training on 18 June here or complete this form to enquire about training or support for your team.
Disclaimer: I used AI to refine the language of this blog. The thinking, examples, and framing are my work.