King's College London

On 25 May 2026, Pope Leo XIV made history by publishing Magnifica Humanitas, the first papal encyclical dedicated to artificial intelligence (AI). The document was signed on the 135th anniversary of Rerum Novarum, a landmark encyclical on labour and the dignity of workers. This timing was no coincidence. Leo's message was stark: the development of AI poses ‘new challenges for the defence of human dignity, justice and labour’, and the control of this technology must not remain in the hands of a few.

Whatever one's faith, the encyclical captures something that many organisations are only now beginning to appreciate. The question of how we govern artificial intelligence isn’t merely a technical matter for engineers and data scientists. Rather, it is a question about humanity. How do we ensure that machines built by humans remain answerable to human values, human oversight and human institutions?

That question is now confronting boardrooms, risk committees and compliance teams with a sense of urgency that would have been difficult to imagine just a few years ago.

From Novelty to Infrastructure

The release of ChatGPT in late 2022 marked a turning point in the workplace. Within months, large language models (LLMs) moved from research curiosity to operational reality across virtually every sector. Law firms, hospitals, financial institutions, government departments and multinational corporations began deploying AI not as an experimental technology, but as core operational infrastructure used to draft documents, process claims, analyse data and support day-to-day decision-making.

"We have moved with extraordinary speed from AI as a concept to AI as a business-critical system,"  says Professor Dan Hunter, Executive Dean of the Dickson Poon School of Law at King's College London. "The challenge now is that organisations have adopted these tools faster than they have developed the frameworks to govern them. That gap is where legal, reputational and operational risk quietly accumulates."

That governance gap is the central challenge of our moment. The OECD has documented it extensively, noting that while policy frameworks are proliferating at national and supranational levels, implementation at the level of the individual organisation remains patchy and inconsistent. The EU AI Act, now in force, imposes legal obligations that many organisations are still scrambling to understand. In the UK, the government's pro-innovation approach has placed even greater responsibility on individual sectors and institutions to self-govern effectively.

 

The New Landscape of Risk

What makes AI governance particularly challenging is that it demands a level of understanding that most organisations have not previously needed to develop. The risks extend far beyond the technical, encompassing strategic, legal, ethical, reputational and operational concerns.

Consider the decisions now being made at every level of an organisation. A procurement team evaluating an AI vendor must now ask questions that would once have belonged to a specialist legal or IT function: Does this model process personal data in a way that is compliant with applicable regulation? What security standards govern the training data? Is there an audit trail that would satisfy a regulator?

Senior leadership faces a different but equally demanding challenge. Boards must determine which AI systems warrant their direct oversight, define and communicate risk appetite, and establish clear lines of accountability when an AI-assisted decision causes harm. The margin between reckless adoption and paralysed inaction is narrow and difficult to hold.

In this complex and sometimes confusing new workplace landscape, one of the most important contributions that serious AI governance education can make is to disrupt a false dichotomy that has taken root in many organisations: the idea that one must choose between enthusiastic AI adoption and cautious risk avoidance. The reality, as both practitioners and policymakers have come to recognise, is more nuanced. The OECD's work on responsible AI makes clear that the goal of governance is not to slow innovation but to make it sustainable by building the trust, accountability and oversight mechanisms that allow organisations to deploy AI confidently and at scale (rather than lurching between uncritical adoption and reactive retrenchment).

The organisations that will navigate the AI era most successfully will not be those that adopted the technology fastest, nor those that held back out of caution. They will be the ones that invested in the governance capacity to deploy AI responsibly, to understand and manage its risks, and to build the institutional trust that sustainable AI adoption requires.

 

What We’re Doing at King’s

This is precisely the spirit that animates the AI Governance Leadership Programme offered through The Dickson Poon School of Law. Designed by Professor (Hon.) Harry Borovick and Professor Dan Hunter in collaboration with leading industry experts, the programme is built around a deceptively simple premise: AI governance is, ultimately, about protecting people. Like the Magnifica Humanitas, the underlying imperative is to ensure that the deployment of powerful technology serves human ends, rather than displacing human agency or eroding human accountability.

King's AI Governance programme is aimed at the professionals who find themselves at the front line of AI adoption: risk and compliance officers, senior managers, governance committee members, general counsel and board-level decision-makers across regulated and high-impact sectors, including finance, law, audit, manufacturing, technology and banking. These are people who must now take decisions about AI that have no real precedent in their professional training. They did not sign up to be AI specialists, but they can no longer avoid the responsibility.