The business case for AI ethics, updated

I first wrote about the business case for AI ethics back in August. I argued that ethics isn't a barrier to innovation — it's a catalyst for better decision-making. Transparency, inclusivity, holistic thinking, playing the long game.

It was a philosophical argument. I believed it, but I didn't have the hard evidence.

So I went looking for it.

I've spent the last few months tracking down research, compiling data and documenting failures. Real numbers. Real costs. Real companies with their names attached to real (pretty embarrassing) consequences. 

If "sort out AI ethics" has been languishing on your to-do list all year and you're doing Q1 planning right now, this is why it needs to move from good intentions to top priority.

The ROI is measurable

The question everyone asks: "But what's the business case?"

Research has found that organisations spending more than 10% of their AI budgets on ethics saw 30% higher operating profit from AI than those spending 5% or less. That's from IBM's Institute for Business Value study, and the performance gap has held for two years.

30% higher operating profit.

Is the ethics spending directly causing those returns? Probably not on its own. The researchers suggest that continued investment in ethics is a reflection of broader AI capability maturity. Companies serious enough to invest in responsible AI are also the ones deploying it most effectively.

There's a quantified cost to getting it wrong

Zillow's AI algorithm overestimated home values, leading to millions in losses, 25% workforce layoffs and closure of an entire division. Air Canada had to pay damages after its chatbot gave a passenger incorrect bereavement fare information. McDonald's shut down its AI drive-through after it kept adding 260 McNuggets to orders. Amazon was fined nearly $900 million in 2021 by the EU, and Meta was fined over $1 billion in 2022 by Ireland for improperly handling personal data.

You might think "that's Big Tech problems, not mine." Maybe. But the regulations don't exempt you because you're smaller, and you've likely got less cushion to absorb the hit.

Regulations have arrived (with more on the way)

The EU AI Act entered into force on 1 August 2024.

February 2025 brought the first major compliance deadline, with prohibitions on certain AI practices now enforceable. August 2025 brought obligations for general-purpose AI models. The high-risk system requirements hit in August 2026.

If you're thinking "I'm not in the EU, doesn't affect me" — maybe reconsider. Planning to sell to European customers? Hire European talent? Work with European partners? It affects you.

Even if you're purely domestic, the EU AI Act is setting the template. The UK government announced in June 2025 that its first UK AI Bill will be introduced in the second half of 2026, and will likely be influenced by the approach taken in the EU AI Act. "We'll think about regulation later" is no longer a valid position to take (I mean, it never was, but now it's wilfully reckless).

Awareness is high (which is good news!)

Employees in many settings are pushing for AI systems that prioritise fairness and bias mitigation, including expectations for data privacy in AI-driven performance evaluations and inclusive design that doesn't exacerbate inequalities.

82% of business leaders say reputational damage is their top concern if ethical standards aren't followed for emerging technologies.

People are informed. They know the risks exist. They're paying attention, and that awareness creates expectations. Trust matters in two directions: your team is watching how you handle AI, your customers are watching too. Both are important. 

The philosophy was sound, the evidence backs it up

When I wrote that August post, I was arguing that AI ethics leads to better decision-making: more transparent, more inclusive, more holistic, more long-term thinking. Turns out that better (+ more responsible) decision-making leads to better financial outcomes. Who knew? (I suspected. Now we have proof.)

If AI ethics is a Q1 2026 priority, there are two ways I can help:

• For solo business owners and freelancers: my AI ethics for freelancers cohort course runs quarterly (next cohort starts 12/01/2026)
• For teams who need to develop AI ethics principles together: I run workshops that take you through the process