Jump to:
On 15 May 2026, Pope Leo XIV signed his first encyclical, Magnifica Humanitas, “on the defence of the human being in the age of artificial intelligence”. The document is the first of its kind dedicated to digital technology and Artificial Intelligence, and was launched at an event alongside Anthropic co-founder Christopher Olah.
The significance and symbolism of the event and its content are all the more striking given that the launch takes place exactly 135 years after Rerum Novarum, the document with which Pope Leo XIII cemented the Catholic Church’s social doctrine in the context of the Industrial Revolution.
Rerum
Novarum
1891
MagnificaHumanitas
2026
It is easy to understand that each element of the new papal document can be interpreted in a personal key and may resonate differently from individual to individual, much like faith and spirituality themselves.
In our interpretation, and beyond the theological language, Magnifica Humanitas is also about different civilisational choices and visions in the context of artificial intelligence, inviting us to answer the question:
Towards what kind of society do we want to build Romania in the years ahead?
In this sense, we also read the papal text as an argument about governance, directly linked to the future of Romanian society: about who decides, who is accountable, and for whom technology is built and used. From this angle, the encyclical says things about Romania that we emphasise at every opportunity and that must be addressed by the Romanian state.
Thus, in the paragraphs that follow, we argue that in the AI era, human dignity is protected through governance, and that for Romania, the governance vacuum (and therefore the absence of vision) is a choice that may come with a price tag translated into an undesirable society.
1. Technology is biased
The encyclical’s starting point is that technology is “never neutral, because it bears the characteristics of those who conceive it, finance it, regulate it and use it”, and that the power that today conceives and finances AI is “predominantly private”, belonging to transnational actors whose resources and capacity “exceed those of many governments”. Thus, when control over platforms, data, and computing power “does not rest with states, but with major economic and technological actors”, that power “tends to become opaque and to evade public oversight”.
For Romania, this threat is abundantly clear: where the state has no strategic and institutional architecture, no state vision, and where “the national AI agenda is written by no one, it ends up being written implicitly by platform providers”. With only 3.07% of companies using AI and without a responsible institution to set direction, Romania remains overexposed, unprepared, absent, the precise secular translation of the papal diagnosis.
Thus, what we have labelled “absence is a choice”, the Pope calls “the abdication of political responsibility”. What we describe as an institutional vacuum, the encyclical describes as a space that the “technological paradigm” inevitably occupies. When efficiency, control, and profit are left unchecked, they “end up dictating what matters and what can be discarded”, all the more so given that Romania’s governance deficiencies are structural.
Similarly, the five principles of social doctrine, the dignity of the person, the common good, subsidiarity, solidarity, and social justice, provide precisely the conceptual framework for Romania’s future in the AI era, as well as the normative framework for what we have proposed on economic grounds: an accountable authority, innovative approaches to regulation, and social criteria for the labour market transition, as a minimum for directing new technologies towards the whole population.
2. Subsidiarity (also) through a sandbox
“Communities and intermediate organisations must not be reduced to mere passive recipients of decisions made elsewhere; they must be able to contribute to discernment and oversight […] to speak of subsidiarity means protecting the capacity of communities to make choices and corrections, not confining them to simple oversight after standards have been fixed elsewhere.”
These passages from the papal encyclical are, in the language of social doctrine, also part of the definition of a regulatory sandbox, a controlled environment in which private actors, public institutions, civil society, and academia test AI solutions before the technology becomes widely deployed and before rules become set in stone.
We have already argued for the necessity of this instrument in the paper “Romania urgently needs a sandbox in public policy“, and the Encyclical adds a moral and normative legitimacy to a pursuit for which Romania has, moreover, a legal obligation from the European Union.
Meanwhile, Romania’s innovation testing ecosystem stands at 12.3% of the European average for product innovation among SMEs, compared to 126% in Estonia, according to the ANIS report, which we have also analysed. In other words, we have both the legal obligation, the economic need, and now the moral argument to innovate and regulate in tandem.
It is worth noting that the papal text goes far beyond the false dilemma that sometimes blocks the Romanian debate, that “if we regulate, we stifle innovation”. The answer is that dignity criteria “do not constitute a brake on innovation; on the contrary, they make it civilised and human”, through “institutions capable of regulating without suffocating and protecting without confiscating”.
3. Who is accountable?
The second backbone of the encyclical is accountability. For AI to serve the common good, “accountability must be clearly defined at every stage”, which means “the possibility of identifying who must answer for decisions, justify them, monitor them, and, when necessary, contest them and repair the damage”, with the caveat that “it is not enough to invoke ethics in the abstract; solid legal frameworks are necessary, along with independent oversight, informed users, and a political system that does not abdicate its responsibility”.
In this regard, in our analysis that also covers the National Strategy for AI 2024–2027, we diagnosed a document “aligned with European standards, but without an explicitly allocated budget and without an institutional structure with a clear implementation mandate”. In short, “a strategy for which no one is explicitly accountable”. It is also worth noting that the document envisaged a dedicated authority for AI governance that was never actually established, in a context where the strategy itself expires next year. This is a faithful illustration of a variant of abdication of the type the Pope describes: ethics invoked, strategy duly created, declarations laden with buzzwords, but without a corresponding legal framework, without oversight, and above all without “someone to answer for it”.
We have dissected this structural problem in depth in the Digital Governance Framework in Romania produced together with Estonian experts from Digital Nation. The discontinuity at every electoral cycle, with three ministers of digitalisation in a single year, condemns any strategy to being a mere document without an owner. The solution we propose is a governance model with a single point of accountability: a Government Chief Information Officer’s office under a Deputy Prime Minister dedicated to digitalisation, and an AI adviser at the Presidential Administration with a mandate extending beyond the governmental cycle.
The encyclical provides the criterion by which any organisational chart should be judged: can the citizen affected by an algorithmic decision, whether regarding access to credit, a public service, or a job, identify who is accountable, contest the decision, and obtain redress? In Romania today, the answer remains no, and the problem extends further into what the Pope describes as the “exclusion of the vulnerable [which] is clothed in a layer of neutrality and objectivity”.
4. The dignity of work
Particularly relevant for Romania is the chapter of the encyclical that addresses work, where the Pope treats mass unemployment as “a true social calamity” that “calls especially upon the state to exercise its responsibility”.
The solution includes public policies that must establish “social criteria for innovation”, so that “every introduction of automation and AI is accompanied by verifiable measures for job protection, retraining, and worker participation”, and that “the cost of adaptation does not fall exclusively on the individual”.
This diagnosis maps onto the double trap in which Romania simultaneously finds itself among the most exposed to automation and among the slowest to adopt it; losing jobs without gaining productivity: between 1.2 and 1.8 million employees fall into high-risk categories of automation in the 2025–2030 horizon, one of the highest exposures in the region.
The institutional response remains, for now, absent: missing are a map of vulnerable jobs, a retraining strategy, a dedicated transition budget, and a National AI Observatory, as proposed by Alexandru Dănescu and Roland Kristo in the Edge Institute report The Impact of Emerging Technologies on Romania’s Labour Market.
Thus, the encyclical adds the moral criterion so often absent from economic calculations: “the human person is an end, not a means, and the economic order must remain subordinate to human dignity and the common good”. A transition that “systematically sacrifices jobs” for profit becomes, beyond a public policy error, an injustice. Automation must therefore “free up human time and capacities, rather than produce exclusion”, which requires proactive, accessible continuing education policies for all.
5. The tension we must acknowledge honestly
It would be dishonest to present the encyclical as an unconditional ally of the pro-adoption agenda. Magnifica Humanitas argues, in several places, for caution and even for “a slower pace in adopting AI”, for the courage to “slow down when everything is accelerating”.
At Edge Institute, we argue to the contrary: that every quarter of hesitation deepens the traps facing Romania, and we advocate for innovation and responsibility, aspects that are not mutually exclusive.
The contradiction is therefore only apparent. Where the Pope criticises uncritical adoption, without safeguards, and at individuals’ expense, we criticise the absence of adoption alongside the absence of safeguards, both to the detriment of (and at the expense of) citizens, at least as long as technology is inevitable. In this sense, the common denominator is that the political factor must assume responsibility.
For Romania, at this moment, the risk is not technological over-optimisation, but becoming the object of technology. In this context, the papal call to “slow down” translates, for us, into “first build the framework”.
Conclusion: dignity is built, not waited for
If we were to summarise the convergence between the Pope’s message and our proposals for the authorities, it would be precisely the papal quote explored above: the need to “regulate without suffocating and protect without confiscating”. We reiterate that we need an active, accountable, and participatory governance, equidistant between the laissez-faire that leaves the vacuum to the platforms and the regulation that frightens innovators.
Translated into three decisions we have formulated in the Call for Romania’s Digital Transformation addressed to the incoming Government, the encyclical’s message for Romania can also be rendered as follows:
- First, a responsible party: a single point of accountability, in the form of a Government CIO under a Deputy Prime Minister for digitalisation, paired with a presidential mandate on AI. This measure must be complementary to the reform of the Authority for Romania’s Digitalisation (ADR), which should develop and guide AI governance.
- Then, an operational sandbox by August 2027, beyond the AI Act obligation, as the institutional embodiment of subsidiarity: the space in which communities actively contribute to innovations and their governance. Likewise, ANCOM, as the designated supervisory authority for the AI Act, must be an additional resource for the compliance function.
- And finally, social criteria for innovation: a map of exposed jobs, a dedicated observatory, and accessible retraining programmes, so that “the cost of adaptation does not fall exclusively on the individual”. Importantly, political and institutional leadership, both from the Government through the Deputy Prime Minister and from the State Counsellor appointed by the President, must align towards directing innovations in the right direction.
In conclusion, Magnifica Humanitas reminds us that technology takes the face of those who govern it or, in their absence, the face of those who profit from the absence of governance. For Romania, today’s vacuum is itself a decision, made by not making it.
And the human dignity that the encyclical places at the centre is defended, concretely, by building the institutions that make the transition a choice we own, rather than a product delivered by others. These conclusions can guide our direction towards the kind of society we want to build for Romania’s future, in the context of artificial intelligence.