THE ETHICS ILLUSION

Why AI Ethics Isn’t Working – And What To Do About It

Over 200 AI ethics guidelines have been published around the world [1]. Companies have spent billions on ethics boards and frameworks. Yet the harms keep coming. Something is badly wrong.

The Problem: All Talk, Little Action

When philosopher Thomas Metzinger helped write the EU’s AI ethics guidelines, he saw something troubling. Industry lobbyists watered down every firm commitment. ‘Red lines’ – things that should never be done – became ‘concerns.’ The term ‘non-negotiable’ was deleted entirely [2]. He called it ethicswashing: talking about ethics to look good, not to change behaviour.

This pattern is everywhere. Companies publish fine words about ‘fairness’ and ‘transparency’ whilst their algorithms discriminate and their systems stay hidden. Ethics boards get set up, then ignored or shut down when they become awkward [3]. Looking ethical has become a substitute for being ethical.

Three Big Problems

1. Principles without teeth. Unlike doctors, AI developers have no shared duty to anyone. There are no professional standards, no licensing, no real consequences for getting it wrong. Vague words like ‘fairness’ can mean whatever suits you. As Oxford’s Brent Mittelstadt put it: ‘Principles alone cannot guarantee ethical AI.’ [4]
2. Rules that can’t keep up. The EU AI Act is a step forward, but technology moves faster than law [5]. The UK’s ‘pro-innovation’ approach relies on goodwill [6]. Neither tackles the basic power gap between tech giants and the people their systems affect.
3. Treating ethics like a bug to fix. Too much AI ethics treats moral questions as technical problems. But fairness isn’t something you can code. Questions about power, justice, and whose interests matter cannot be solved with a checklist [7]. And focusing on science-fiction scenarios about robot overlords distracts from the real harms happening now [8].

Why This Matters

AI systems are already deciding who gets loans, jobs, housing, healthcare, and prison sentences. The damage is real and it falls hardest on those already disadvantaged. A 2024 survey found that 73% of senior executives think AI ethics guidelines matter – but only 6% have actually created any [9]. That gap between words and action is the problem in miniature.

What Needs To Change

For business leaders: Stop treating published principles as proof you’re ethical. Give ethics teams real power to stop projects and raise concerns [3]. Listen to the people your systems affect, not just your shareholders. And get ready for proper regulation – it’s coming.

For regulators: Don’t let industry dominate your expert groups [2]. Focus on enforcing rules, not just writing guidance – consequences matter more than codes. Help turn vague principles into practical standards. And work with other countries to stop companies shopping around for the weakest rules.

The Key Takeaway

The question isn’t whether AI needs ethics. It’s whether what we’ve got actually works. Right now, it doesn’t. We need less talk and more action: better institutions, real accountability, and honesty about what’s going wrong.

References

[1] Corrêa, N.K., et al. (2023). ‘Worldwide AI ethics: A review of 200 guidelines.’ Patterns.
 doi.org/10.1016/j.patter.2023.100857

[2] Metzinger, T. (2019). ‘Ethics washing made in Europe.’ Der Tagesspiegel.
https://www.tagesspiegel.de/politik/ethics-washing-made-in-europe-5937028.html

[3] Tsamados, A., et al. (2024). ‘Ethics Washing, Symbolic Ethics Offices, and AI Systems.’ Canadian Journal of Philosophy.
cambridge.org/core/journals/canadian-journal-of-philosophy

[4] Mittelstadt, B. (2019). ‘Principles alone cannot guarantee ethical AI.’ Nature Machine Intelligence.
doi.org/10.1038/s42256-019-0114-4

[5] Mahmutovic, A. (2025). ‘The EU AI Act: a proactive framework.’ Int. Journal of Law and Information Technology.
doi.org/10.1093/ijlit/eaaf028

[6] UK Dept. for Science, Innovation and Technology. (2023). ‘AI regulation: a pro-innovation approach.’
gov.uk/government/publications/ai-regulation-a-pro-innovation-approach

[7] Schultz, M.D., et al. (2024). ‘Digital ethicswashing: a systematic review.’ AI and Ethics.
link.springer.com/article/10.1007/s43681-024-00430-9

[8] Asaro, P.M. (2024). ‘The Rise of AI Ethics.’ Social Research / Public Seminar.
publicseminar.org/2024/02/the-rise-of-ai-ethics

[9] World Economic Forum. (2024). ‘Why corporate integrity is key to shaping future use of AI.’
weforum.org/stories/2024/10/corporate-integrity-future-ai-regulation

The Full Report

This summary is drawn from the full report:

A Philosophical Critique of AI Ethics: The Gap Between What Companies Say and What They Do

To request a copy of the full report, please send a message via the contact page (including your email address) and I will reply with a pdf attachment.