Ethical AI Isn't a Pipe Dream. It's Already Helping UK Charities Today
- 11 minutes ago
- 4 min read
A SherlockAI case study
Everyone's talking about ethical AI. Fewer people are actually building it. At SherlockAI, we're a Community Interest Company (CIC), so "ethical" isn't a buzzword we bolt onto a sales page, it's the whole reason we exist. This case study is about one project that proves it: a real, working, in-use solution for a major UK charity, built the ethical way, from the ground up.
We've kept the charity's name out of this post, because respecting the privacy of the people and organisations we work with is part of what "ethical" means to us. But everything below is real, and it's live right now.
The Challenge: 24/7 Multilingual Support, Without Compromising on Values
Our client is a well-known UK charity that actively reaches out to support people in need. They wanted to go beyond their existing helpline and website, and offer anonymous chat support, in multiple languages, around the clock.
On paper, that sounds like a job for any off-the-shelf AI chatbot. In practice, this charity wasn't comfortable with that at all. And for good reason.
Why "Just Use ChatGPT" Wasn't Good Enough
Three things stood in the way of a standard AI chatbot:
Data privacy. They didn't want confidential, sensitive information about vulnerable people being sent off to big US AI tech giants.
Accuracy. Even the most advanced mainstream AI tools can confidently make things up — what the industry calls "hallucination." When someone reaches out for support in a crisis, a confidently wrong answer isn't just unhelpful, it's dangerous. This charity cared too much about the people they serve to accept anything less than fully honest, correct responses.
Environmental impact. As an organisation that's ethical in everything they do, they didn't want to contribute to the excesses of big data centres either.
That's when SherlockAI stepped in.
Our Solution: A Bespoke, Ethically-Built AI Support Tool
Working side by side with the charity's own team, we built a one-of-a-kind, bespoke support tool for their service users — not an off-the-shelf product with the logo swapped out.
We coupled an ethically trained AI from Apertus with the charity's own knowledge and expertise, so it could only draw on information the charity itself had verified. And crucially, if the tool was ever asked something it couldn't answer with complete certainty, it wouldn't guess. It would simply say "I don't know" — and point the person straight to a helpline number.
No hallucinations. No made-up answers. No pretending to know something it didn't. Just honest, accurate support, available in a dozen languages, 24 hours a day.
Built on Truly Ethical Foundations
Every part of the stack was chosen with ethics in mind, not just cost or convenience:
UK, renewable-powered hosting. The solution runs on a UK cloud host, The Green Hosting Co., powered entirely by renewable energy.
Hydro-powered, privacy-first AI. We partnered with an ethical AI LLM supplier based in Switzerland, Schatzai, who offer exceptional levels of data security and privacy, run entirely on hydro-electric power, and provide access to privately hosted, ethically trained AI models.
That means every query, every response, and every watt of energy behind this tool sits on ethical ground — not just the parts people can see.
Security You Can Actually Trust
When you're supporting people in vulnerable situations, security isn't optional — it's the whole job. We built multiple layers of protection into the solution end to end, including:
A detailed Data Protection Impact Assessment (DPIA)
An independent external security audit, carried out by Risk Crew
We won't compromise on the personal data of people who are already in need. Full stop.
Charity-Owned. Charity-Controlled. Fully Independent.
Here's the part we're proudest of: the charity owns the solution.
We built them the ability to upload and update their own information, so their service users always get the best, most current help available — across a dozen languages. They choose and control their own suppliers and hosting providers. There's no lock-in, no dependency on us, no fine print. They're completely independent.
Why This Matters for Charities, CICs and Social Enterprises
We're a CIC too. We know exactly how hard it is to get meaningful things done on a minimal budget, with deep ethical guidelines, while working with people who really need your help. That's precisely why we build things with organisations, in a way that works for them — not a one-size-fits-all product forced to fit.
Ethical AI, done properly, isn't slower, softer, or less capable. It's more trustworthy, more accountable, and — as this project shows — it works. Right now. Today.
Let's Talk
If any of this resonates with your organisation, or you're just curious about what ethical AI could look like for the people you support, please feel free to reach out. We're genuinely excited about where ethical AI is going, and we'd love to help you make it happen.
SherlockAI is a Community Interest Company helping charities, social enterprises and third-sector organisations build AI solutions that put ethics first — without compromise.