THE SUMMER of 2026 has been dominated by growing public concern about the impact of AI technologies on our societies, and the sense that we are now in an AI race equivalent to the arms race. The recent admissions by the leadership at Anthropic and Open AI that there is a risk of human extinction if the AI race is not constrained2 confirms what advocates of ethical AI have been insisting on for a decade, namely that governments have a duty to regulate AI. Until recently however many citizens have been unaware of the depth of the impact of AI on our societies, and of the ethical challenges that they pose. Some commentators warn of a dystopian future, with the displacement of humans by superintelligences and deepening polarisation and inequality. Others anticipate a future of greater wealth and opportunity and of significant scientific advances. Whatever the ultimate outcome, the accelerating developments in AI will have a fundamental impact on all aspects of our lives and are likely to transform our political and social institutions. As AI continues to be deployed in policing, healthcare, justice, education, business, media and the military it raises new ethical questions about how we should act in the world. It also re-frames current ethical questions in a new and more momentous context. The Christian tradition, in its manifold expressions has much to contribute to the societal conversation, particularly with its insistence that this AI revolution must respect the requirements of human dignity, and the common good. My title “ethical by design” echoes the phrase “privacy by design,” an approach to technology design that is a legal requirement in the European Union. Privacy by design requires the principle of data privacy to be embedded in technology from the beginning of the design process all the way through to its deployment. In this framework privacy is the default, not an afterthought. Its premise is that technology design is not value-free but rather reflects, embeds and amplifies specific values (for example efficiency or accuracy or equality or transparency). It operates on the assumption that these embedded values not only influence how we interact with the technologies, but that they also affect our identities and our relationships, and when deployed at scale, can have significant political and social impact. An “ethical by design” approach therefore insists that ethics must be the default, not an afterthought. It requires the development of AI systems to respect the dignity of each person, the principles of social justice, the parameters of ecological responsibility and our creaturely status. Artificial Intelligence: the ethical challengesNotwithstanding the significant benefits recognised to accrue from AI, questions about the ethical parameters of its use are important for society and have been recognised as such by ethicists from different traditions for more than a decade. Their discussions initially focused on the technology itself, particularly on questions about the nature and impacts of generative large language models (LLMs) and their emergent capabilities. They quickly developed to include the multiple domains of AI deployment including biomedical ethics, war ethics, social and political ethics. Underpinning these has been the crucial ethical task of establishing the upper limits of AI development. Fundamental ethical concerns have focused on the accuracy and reliability of the data on which the LLMs are trained, the values encoded in the design of algorithms, the risks of amplifying and reinscribing racial, ethnic and gender-based biases through the use of predictive technology, and the dangers of delegating moral decision-making to algorithms. Since AI technologies are now deployed in almost every domain of life, including education, business, finance, welfare, healthcare, policing, the administration of justice, as well as border control and warfare, these algorithmic systems have far-reaching impacts for individuals and communities. Among the ethical problems they can give rise to are “inconclusive evidence leading to unjustified action, inscrutable evidence leading to opacity, misguided evidence leading to unwanted bias and unfair outcomes leading to discrimination.”3 Moreover as is evident from analyses of the deployment of algorithmic systems in migration and border control,4 in the administration of criminal justice,5 media,6 and many other domains, when any of the problems noted above are encoded in such systems, the impacts can be damaging and unjust. The fact that AI technologies are embedded across many of our critical infrastructures also means that multiple inequalities can fundamentally impact the services on which our societies depend and gives rise to the worry of infrastructure takeover by AI hackers. In some cases, algorithmic systems replicate existing biases and injustices, in others they magnify them, while in others still they create new forms of exclusion, including through their intersectional impacts. Moreover, as argued by Abeba Birhane this “mathematisation and formalisation of social issues brings with it a veneer of objectivity and positions its operations as value-free, neutral, and amoral.”7 This serves to obscure and justify discriminatory outcomes, and, as Birhane notes, leaves unaddressed the problem of “the mathematisation of ambiguous and contingent issues, historical inequalities, and asymmetrical power hierarchies or unexamined problematic assumptions that infiltrate data practices.”8 When deployed at scale and obliquely, evaluated according to opaque metrics) and managed by unaccountable authorities, such algorithmic systems raise systemic ethical challenges for the well-being of individuals and societies. Nor are AI systems developed or deployed in a political vacuum. Society’s values and power structures shape how technology is used and raises concerns, about potential threats to human rights and civil liberties, about new and more consequential forms of the digital divide and global inequality, and about the ecological impact of these resource-hungry technologies. Clarity about the ethical principles that ought to guide the design and deployment of the technology is therefore essential. Principles for AI Ethics and the risks of ‘ethics washing’In fact, the last decade has seen a lively discussion about the ethics of AI. There has been a multitude of industry standards focusing on ethics which have sought to identify the principles that should guide AI development. Ethics guidelines for engineers and developers have been issued by professional bodies...
