9 principles for adopting AI responsibly: Lessons from 17 B Corp leaders

Recent research by the British Chamber of Commerce shows that over half of UK SMEs are actively using AI, with deeper, agentic adoption predicted. Whether it’s automating the heavy lifting, streamlining operations, or helping teams brainstorm, the potential to work smarter is huge.

For purpose-led businesses, however, the conversation goes beyond efficiency. The challenge isn’t whether to adopt AI; it’s how to adopt it in a way that reflects the organisation’s values, protects trust and delivers better outcomes for people, customers and communities.

During B Corp Month, we spoke to 17 B Corp leaders from organisations spanning cybersecurity, venture capital, childrenswear, nature fintech, retail and creative agencies. We wanted to understand how they were approaching AI in their own organisations: not just the tools they were using, but the principles guiding their decisions.

What struck us was that, despite working in very different sectors, many of them were wrestling with the same questions. How do you balance innovation with responsibility? Where does human judgement remain essential? How do you ensure AI strengthens your purpose, rather than distracting from it?

Rather than offering a list of AI tips, this article brings together the recurring themes that emerged from those conversations. These nine principles reflect the practical, thoughtful approaches B Corp leaders are taking as they embrace AI responsibly.

With powerful technology comes a whole new set of responsibilities. Integrating AI means grappling with big questions around privacy, bias, intellectual property, and knowing exactly when human judgement needs to take the wheel.

There is a significant environmental footprint to consider, too. AI models demand massive computing power. To put it in perspective: the International Energy Agency predicts that by 2030, AI data centres will consume nearly 3% of all global electricity. Even more surprisingly, running just 10 to 50 simple AI prompts can use up a 500ml bottle of fresh water for server cooling.

These aren’t reasons to avoid AI, but they are incredibly good reasons to use it with care.

And that was the exact thread running through all 17 of our conversations. The B Corp leaders we spoke to aren’t just blindly adopting new tools; they are making highly conscious decisions. They are actively choosing where AI adds genuine value, where human expertise is irreplaceable, and establishing the governance needed to use AI with confidence.

While every business tackles this a little differently, nine core principles emerged time and again. Think of these less as a rigid checklist, and more as a practical, real-world framework to help you adopt AI in a way that stays true to your values.

AI can deliver enormous value, but that doesn’t mean it’s the right tool for every task. A responsible approach always starts with a simple question: is the outcome really worth the energy, cost, and risk involved?

For Eric Zie, Founder and CEO of GoCodeGreen, this mindset is central to how they operate. GoCodeGreen helps organisations measure and reduce the carbon footprint of their digital products. His team uses AI to analyse software and cloud infrastructure, pinpointing where resources are being wasted so companies can cut both emissions and running costs.

Here, AI isn’t being used just because it’s available; it’s deployed specifically where it creates a clear, net-positive impact that directly serves the company’s purpose.

As Eric puts it:

“Rather than AI becoming just another source of digital emissions, we use it to help reduce them.”

Eric Zie, Founder and CEO, GoCodeGreen

AI is a powerful tool, but being responsible means knowing when a simpler option will do the job just as well.

At nature fintech company CreditNature, Operational Excellence Manager Robyn Silcock relies on a wonderfully straightforward rule with her team: if a standard web search is enough, use that instead. Basic look-ups and routine searches rarely justify the computing power or energy footprint of a generative model.

Ultimately, it comes down to proportionality: save AI for tasks where it adds genuine value – like complex synthesis, deep analysis, or creative brainstorming – and stick to standard tools for everything else.

“We treat AI like any other resource. We use it intentionally, not just because it’s available. Our team has a simple test: if a quick Google will do, then don’t use AI.”

Robyn Silcock, Operational Excellence Manager, CreditNature

One of AI’s greatest strengths is helping us find and organise information. One of its biggest risks, however, is tempting us to accept its outputs as absolute truth.

Donna Okell, CEO of UK for Good, shared a brilliant mental model to keep this in check: treat AI like a librarian, never an oracle.

Where a librarian helps you navigate the shelves, an oracle expects blind faith. Generative tools can (and often do) fabricate citations, misinterpret data, or invent facts out of thin air. Using AI to do the initial research legwork saves massive amounts of time, but source-checking, critical thinking, and final decisions must always stay in human hands.

“Use it as a librarian: ‘Find me five research articles published between these dates, with references.’ Then go and look at those sources yourself and make your own judgement. Anything else, be very, very careful.”

Donna Okell, CEO, UK for Good

Introducing new tech without context breeds natural anxiety. People want to know why a tool is being introduced, how it changes their workflow, and whether it’s replacing human effort.

For James Ewin, Founder at ORCA, tackling those questions head-on is the only way to build lasting trust:

“It’s about making sure that there are clear guardrails and the team really understand how to use it, when to use it, and when not to use it… I want to support my team and ensure that we’re using AI to benefit us and not just to add further complication to the process.”

James Ewin, Founder, ORCA

The takeaway: Explaining the “why” to your team starts with asking the right questions at the leadership level:

1. The Problem: What specific challenge are we solving?
2. The Value: Where does AI truly add value over existing methods?
3. The Collaboration: How will people and AI work side by side?
4. The Safeguards: What human oversight is required?
5. The Alignment: How does this reflect our organisation’s values?

By making conscious, transparent choices, purpose-led businesses can explore new capabilities while keeping their teams aligned and confident.

The strongest theme across all 17 interviews was clear: AI should elevate human expertise, not replace it. True oversight isn’t just proofreading; it’s building thoughtful controls into every step of the process.

It’s something we’re conscious of, too. Even in our automated translation workflows, skilled linguists are involved in the set-up, testing and refining of the AI models. Without that step, the quality of the output would be compromised.

At sustainability consultancy AG Impact, Communications Manager and DEI Lead Emily Cusick uses a crisp framework to safeguard quality and trust:

“Four eyes, two rounds. Transparency of use is key to our values, and so is human oversight to ensure accuracy. Human judgement is still critical in producing authentic communications and building trust.”

Emily Cusick, Communications Manager and DEI Lead, AG Impact

The takeaway: Under Emily’s rule, high-stakes or external content must pass through two human reviewers before publication. But effective oversight also means asking smart questions before hitting generate:

1. Tool selection: Is AI appropriate here, and are we using the right model?
2. Context: Does the prompt include the right source data and brand voice guidelines?
3. Risk assessment: What level of human scrutiny does this specific output need?

AI can help talented teams work faster, but critical thinking, brand alignment, and final accountability must stay firmly in human hands.

Most organisations wouldn’t sign a contract with a new supplier without checking their credentials, security, and environmental track record. AI software shouldn’t get a free pass just because it’s shiny, exciting, or easy to sign up for.

It’s easy to get drawn in by flashy features and promises of massive productivity gains. But purpose-led organisations look well beyond raw functionality. They ask how the model was trained, where their data actually goes, and whether the provider’s practices align with their own ethics.

One example we heard came from brand and impact agency Civilian, who described evaluating potential AI tools against four key criteria: privacy, security, transparency and environmental impact.

Ultimately, choosing an AI vendor isn’t just an IT decision — it’s a trust decision. Giving AI tools the same procurement rigour as any other vendor ensures you don’t inadvertently compromise your values for the sake of convenience.

“We actively research which AI companies are taking the most ethical and transparent approach, and we evaluate the environmental footprint.”

Travis Kushner, Account Director and Director of Business Development, Civilian

Because AI models are trained on human data, they inevitably carry human flaws, stereotypes, and historical patterns. Responsible organisations don’t expect AI to be objective; they build workflows that actively search for its blind spots.

Kelly Smith, Strategic and Creative Partner at NEO, highlights that tackling bias starts with a fundamental shift in mindset:

“You have to assume there is bias somewhere in the system. The question isn’t whether it exists, it’s whether you’ve designed your processes to recognise and challenge it.”

Kelly Smith, Strategic and Creative Partner, NEO

Research consistently shows that AI trained on unrepresentative data can quietly reinforce inequalities. From subtly exclusionary marketing copy to biased screening filters, even seemingly neutral outputs reflect the assumptions baked into their training data.

The takeaway: To catch these blind spots before content goes live, teams should routinely ask:

• Inclusion: Could this output unintentionally exclude or disadvantage anyone?
• Perspective: Have we considered viewpoints outside our own team or demographic?
• Empathy: Are we asking AI to make decisions or write content that demands genuine human empathy?

Managing bias isn’t a one-off setting you toggle on in software; it’s an ongoing habit that requires diverse teams, open dialogue, and a healthy dose of scepticism.

Responsible AI isn’t just about deciding where to use the technology; it’s also about deciding where not to.

The organisations we spoke to recognised that some tasks are simply too important, too sensitive or too closely tied to human relationships to be handed over entirely to AI. Establishing those boundaries helps ensure technology supports, rather than undermines, trust.

For sustainable clothing brand Y.O.U Underwear, that’s image generation, and for two reasons: it’s considerably more energy-intensive than text, and airbrushing or manipulating images runs directly counter to a brand built on showing real bodies honestly. So it’s simply off the table, plain and simple.

“We’re not using it for image generation. It doesn’t align with our values about not manipulating images.” 

Sarah Jordan, Founder and CEO, Y.O.U Underwear

Every organisation’s boundaries will be different. Some may choose not to use AI for recruitment decisions, performance reviews or sensitive customer communications. Others may require human approval for legal advice, financial decisions or external communications.

The important point is to make these decisions deliberately, rather than allowing AI to become embedded in areas where human judgement, empathy or authenticity are essential.

As AI capabilities continue to evolve, organisations should regularly revisit these boundaries. What matters most is not having a long list of restrictions, but having clear principles that reflect your organisation’s purpose, values and responsibilities.

Saving time is one of AI’s biggest selling points. But the real question for purpose-led businesses is: what do you actually do with those saved hours?

Rather than measuring success purely by speed or volume, the leaders we spoke to were deliberate about where that reclaimed energy goes. By handing off routine admin and repetitive tasks to AI, teams free up valuable bandwidth for deeper strategic thinking, creative problem-solving, and genuine human connection.

Allisha Heidt, Founder and CEO of zero-waste eCommerce brand Chickpeace Planet, uses AI selectively for routine tasks specifically to protect time for what matters most:

“If you’re able to buy back your time, you can take those resources and give them back into your team or your community.”

Allisha Heidt, Founder and CEO, Chickpeace

Used thoughtfully, AI doesn’t just make us more productive. It gives us the space to be more creative, more present, and more human.

No organisation has all the answers when it comes to AI.

The leaders we spoke to came from different sectors, faced different challenges and were at different stages of their AI journey. Yet there was remarkable agreement on one point: responsible AI isn’t about resisting technology; it’s about using it deliberately, transparently and in ways that strengthen rather than compromise your values.

As AI continues to evolve, so too will the questions organisations need to answer.

• Where does AI genuinely add value?
• Where is human expertise indispensable?
• How do we ensure our governance keeps pace with the technology?
• And how do we make sure the benefits of AI are shared in ways that support our people, our customers and society?

At Comtec, we’ve reached many of the same conclusions through our own work helping organisations create multilingual content responsibly. We believe AI delivers its greatest value when it’s combined with human expertise, clear governance and a deep understanding of context, culture and quality.

The future doesn’t belong to organisations that use the most AI. It belongs to those that use it with the greatest purpose.

If your organisation is exploring how to introduce AI more responsibly, whether that’s creating policies, reviewing existing workflows or deciding where human expertise should remain at the heart of the process, we’d love to share what we’ve learned.

We’re also continuing these conversations with purpose-led organisations across different sectors, and we’d welcome the opportunity to hear your perspective.

Get in touch to discuss how your organisation is approaching responsible AI, or explore how Comtec helps organisations combine AI and human expertise to create multilingual content that is accurate, authentic and trusted.

This article was informed by conversations with 17 B Corp leaders. We would like to thank everyone who generously shared their experiences, ideas and perspectives with us.

GoCodeGreen, CreditNature, UK for Good, CapEQ, AG Impact, Civilian, NEO, Y.O.U Underwear, Chickpeace, Beyond Encryption, Kamwell, Fast Penny Spirits, Mama Bamboo, Hemmings House, BH&P, ORCA and Better Ventures.