What Is AI Ethics, and Why Is Everyone Suddenly Asking?

AI Ethics

A year ago, nobody asked you about AI ethics. Now it turns up everywhere. A tender asks for your "approach to responsible AI". A new client wants to know whether a person checks what your chatbot says. A member of staff asks if it's OK to run CVs through an AI tool. And the term itself gets searched more than a thousand times a month in the UK alone.

Businesses are exploring, adopting and deploying AI tools for all different reasons, from improving productivity and customer engagement to introducing new products and services. However, adopting AI doesn’t mean an organisation has considered the ethical implications of using it.

Maybe the phrase ‘AI ethics’ sounds academic, so a lot of business owners park it or maybe they just don’t even consider it during the decision to adopt it. That's a mistake, because the questions behind it are practical responsibilities. Is the AI treating people fairly? Could the system produce inaccurate outcomes? Do people know they're dealing with AI? Who is accountable for its recommendation? If it gets something wrong, who puts it right?

This guide explains what AI ethics means in plain terms, why it matters when using AI or when introducing into the organisation, why it has moved so fast from conference talks into supplier questionnaires, and what a small or mid-sized UK business can do about it this month.

What AI ethics means, in plain English

AI ethics is the set of principles that guide how AI is built and used, so that it helps people and doesn't harm them. For a business, it comes down to a short list of questions you should be able to answer for every AI tool you use.

Is it fair? AI learns from past data, and past data carries past bias. A recruitment tool trained on ten years of hires can learn to prefer the kind of people you've always hired. Nobody programmed that in. It still happens.

Is it clear? People should know when they're dealing with AI and roughly how it reaches a decision. A chatbot that lets customers think they're talking to a person fails this test straight away.

Is someone accountable? Every AI tool needs a named owner who answers for what it does. "The AI did it" is not a defence, as Air Canada found out (more on that below).

Is personal data respected? AI tools run on data, much of it personal. Ethical use means only using data people would expect you to use, keeping it safe, and not feeding it into tools that keep or reuse it without permission.

Can people challenge it? If an AI-assisted decision affects someone, such as a loan application, a job application or an insurance price, they should be able to ask a person to look at it again.

Does it work safely? The tool should do what you think it does (intended purposes), reliably (conduct appropriate testing), and you should be checking that it still does over time (ongoing monitoring to identify inaccurate, unreliable or unexpected outcomes).

You'll see these ideas written up in different ways by different bodies. The UK government's own approach to AI regulation rests on five principles which is noted in their 2023 AI regulation framework:

  • safety, security and robustness;

  • transparency and explainability;

  • fairness;

  • accountability and governance;

  • and contestability and redress.

UNESCO's Recommendation on the Ethics of AI, adopted by all 193 member states in 2021, and the OECD AI Principles cover the same ground. The wording changes. The questions don't.

These principles provide guidance, but organisations need to determine how they apply to their particular AI systems, business activities and potential impacts.

AI ethics and AI governance are not the same thing

The two terms get used as if they mean the same thing, and it causes confusion.

AI ethics is the what: the principles you want your AI use to meet. AI governance is the how: the policies, owners, checks and records that make sure those principles actually happen day to day, meaning translate those principles into practical processes, responsibilities and human oversight arrangements.

A business can have strong commitments about fairness and still use a biased tool, because nobody was given the job of checking. 

Ethical principles provide direction, but governance establishes how those principles are considered when AI systems are selected, approved, deployed and monitored.

For example, an organisation may be committed to fair recruitment practices. AI governance provides a way to assess recruitment tools, identify potential bias, establish appropriate human oversight and review outcomes.

If you want to see what that looks like in practice, our AI governance services page sets it out.

Why everyone is suddenly asking about AI ethics

The topic has been around for years. Four things have pushed it onto business owners' desks in the last 12 to 18 months.

Why AI ethics is important when business adopts AI

AI ethics is not a new concept. However, as AI tools become more accessible, organisations have more opportunities to introduce AI without necessarily understanding the risks, limitations or wider implications. These can extend beyond regulatory compliance to include loss of customer trust, reputational damage, strained client relationships and potentially lost sales or business opportunities.

The need to consider ethical issues depends on how AI is being used, the decisions it supports and the people who may be affected.

Employees may be using AI without any formal approval

Not all AI use beings with a formal business project. Employees may independently experiment with AI tools, or AI capabilities may be introduced through software already used by the organisation. This then raises questions about whether those tolls have been assessed and approved, what information employees are sharing, how outputs are being checked and who is responsible for oversight.

Customers may not realise they are interacting with AI

As businesses introduce AI-powered chatbots, virtual assistants and automated customer service tools, it can become increasingly difficult for customers to distinguish between communicating with a person and interacting with AI.

When businesses don’t make this clear, customers may feel misled or frustrated, particularly when they are trying to resolve a complaint, purchase goods, obtain accurate information, or want to speak directly with someone who can assist them. The concern isn’t simply that AI is being used. It’s whether customers understand who or what they are interacting with, whether the information they receive is reliable and whether they can reach a person when their inquiry requires human involvement.

Poorly managed AI management that doesn't take AI ethics into account does affect customer trust, satisfaction and loyalty. In some circumstances, they can lead to complaints that cost the organisation time and effort to resolve, lost sales or customers choosing to take their business elsewhere.

This is where AI ethics and governance become really important. Businesses should consider transparency, customer expectations, the limitations of their AI systems and appropriate escalation processes before introducing any AI into customer-facing services.

Suppliers may ask about responsible AI

Some organisations are introducing AI-related questions into supplier assessments and procurement processes. These may cover what AI systems are used, how information is processed, whether AI-related risks have been assessed and who is responsible for approving and overseeing AI use.

Businesses that can explain their AI governance arrangements are in a far better position to respond to these questions and demonstrate how relevant risks are being managed.

Mistakes have become public and expensive

In 2024, a Canadian tribunal ordered Air Canada to honour a refund policy that its chatbot had made up. The airline argued the chatbot was responsible for its own words. The tribunal disagreed. Marnie's article ”The AI Did It” Didn't Save Air Canada unpacks what that means for smaller firms. Cases like this have made "who is accountable for the AI?" a board-level question.

The rules are catching up

The UK doesn't have a single AI law. Existing law still applies to AI, though. UK GDPR covers the personal data AI tools use, the Equality Act covers discrimination by an algorithm just as it does by a person, and consumer law covers what a chatbot promises. The ICO has published detailed guidance on AI and data protection. If you sell into the EU, the EU AI Act adds its own duties, and our EU AI Act guide explains who it reaches.

What AI ethics looks like in a small business

AI Ethics

You don't need an ethics committee. For most businesses with fewer than 250 staff, AI ethics shows up in a handful of everyday decisions. Here are some examples of where ethics considerations may arise:

  • Customer service chatbots. Tell people they're talking to AI (be transparent), stop the bot from making promises it can't keep, and give them an easy route to a person. Marnie's seven checks before adding a chatbot is a good place to start.

  • Recruitment. If AI screens, ranks or accesses job candidates, organisations need to check its results for potential bias and discrimination, also consider data protection requirements, understand how the tool influences decisions and make sure a person makes the final call. Having a person approve and AI recommendation does not automatically make the outcome fair either. Applicable laws still apply with using AI in the recruitment process and failing to comply can result in discrimination claims, regulatory action and potentially significant financial penalties. Under the EU AI Act, certain AI recruitment systems are also classified as high-risk and subject to additional requirements.

  • Marketing and content. Check AI-written claims before they go out, and be careful with AI images of real people. Consider accuracy, intellectual property, privacy and whether AI-generated contact could mislead customers.

  • Pricing, credit and insurance decisions. These affect people's lives directly, so they need careful assessment of potential impacts, appropriate oversight, and processes for addressing concerns or challenges. Additional legal requirements may apply depending on the decision and how AI is used.

  • New AI tools in general. Before a new tool goes live, check what data it uses and whether the law expects a Data Protection Impact Assessment. Also consider its intended purpose, potential business impacts, supplier arrangements, accountability and ongoing monitoring requirements.

Common myths about AI ethics

"It's only for big tech." The firms building AI models carry a lot of the responsibility, but you're responsible for how you use them. If your chatbot misleads a customer, that's on you, not the software or AI model provider.

"Our AI provider handles the ethics." Providers can make a tool safer. They can't decide what data you put in, who you use it on, or whether a person checks the output.

"Ethics slows everything down." A few simple checks at the start are much quicker than fixing a public mistake, a complaint to the ICO or a lost tender.

"We don't really use AI." Most businesses use more AI than they think, built into email, CRM, accounting and design tools. You can't use it responsibly if you don't know where it is.

Five steps to start this month

  1. List the AI you use. Every tool, who uses it, and what data goes into it. Include the AI features built into software you already pay for. Record its intended purposes, the business process it supports and whether it has been approved.

  2. Give each tool an owner. One named person who answers for how it's used and checks it still behaves. The tool owner is responsible for understanding any associated risks, monitoring its performance, and escalating concerns.

  3. Write an AI Use Policy that staff can easily follow. What staff can use, what they can't, and what data must never go into an unapproved AI tool. A page or two is enough to start that reflects the organisation’s activities and be supported by appropriate employee guidance and awareness.

  4. Keep a person in the loop for decisions about people. Hiring, pricing, credit and complaints should never rest on an AI output alone.

  5. Tell people when AI is involved. On your website, in your emails, in your privacy notice and in your chatbot. Being open builds trust faster than anything else on this list. Consider what information individuals should receive about relevant AI use, taking account of the context and applicable transparency obligations.

These steps provide a starting point. As AI use changes or expands, businesses should reassess the associated risks and review whether existing governance arrangements remain appropriate and compliant.

Fair, open and accountable

AI ethics sounds like a big, abstract subject. For a business, it comes down to a few plain questions: is your AI fair, are you open about it, is someone accountable, and can people challenge it? Your buyers, your staff and the law are all starting to ask. The businesses that can answer clearly will win more trust and more work or sales.

At Hyplon, we helps UK businesses turn good intentions about AI into simple, practical governance: an AI inventory, clear policies, named owners and checks that hold up when a buyer or regulator asks. If you'd like a second pair of eyes on how your business uses AI, take a look at our AI governance services or book a 20-minute conversation with Marnie or Roger.

Next
Next

AI Meeting Transcript: Just Because AI Can Record a Meeting Doesn’t Mean It Should