World
Pope Leo Takes Aim at AI
— 3 mins read
The 42,300-word document demands regulation, worker protection, and human control of AI weapons
Pope Leo XIV has published his first encyclical, a 42,300-word warning that artificial intelligence is concentrating power in a small group of companies and the people who fund them. The Vatican released the document, titled Magnifica Humanitas, on 25 May 2026.
An encyclical is a formal letter setting out official Catholic teaching to the worldwide Church. The Pope signed this one on 15 May, 135 years to the day after Rerum Novarum, the 1891 letter in which Leo XIII set out the Church's response to the Industrial Revolution. The Vatican press office confirmed the new document is deliberately positioned as its successor.
What the Pope wants governments to do
The encyclical calls for regulation of AI, protection and retraining for workers whose jobs it replaces, human oversight of AI weapons, and education that teaches people to question what machines produce. It argues AI is not evil in itself, but is not neutral either, according to Vatican News.
'AI tends to amplify the power of those who already possess economic resources, expertise and access to data', the Pope wrote. Small but highly influential groups, he said, 'can shape information and consumption patterns, influence democratic processes and steer economic dynamics to their own advantage'.
The document was presented in Rome alongside Christopher Olah, co-founder of AI firm Anthropic, the New York Times reported. Anthropic is a private company valued at $965 billion (c.£714 billion).
What the machines can already do
The Pope's timing tracks the technology's pace. Research cited across the industry found frontier AI's ability to complete human tasks doubling roughly every seven months, with the computing power used to train top models doubling about every five months since 2020.
Anthropic's Claude Mythos model showed what that means in practice. In controlled tests by Anthropic and the UK's AI Security Institute, the model broke into weakly defended corporate networks end to end in three of ten attempts, The Decoder reported. It also found thousands of software vulnerabilities on its own, according to BBC News. Anthropic has restricted the model to vetted partners.
Ukraine uses AI to guide drones without GPS, recognise targets, and intercept Russian attack drones.
Early commentary on the encyclical claimed Mythos 'can hack any online platform', including national defence and banking systems. That overstates the record: its capabilities were demonstrated in controlled testing, and no real-world attacks on defence or banking infrastructure have been attributed to it.
The technology is already shaping war. Ukraine uses AI to guide drones without GPS, recognise targets, and intercept Russian attack drones, the New York Times reported. The encyclical warns that machine-speed targeting risks removing human judgment from decisions to kill.
What Westminster announced the same week
Four days before the encyclical appeared, Chancellor Rachel Reeves delivered an emergency fiscal statement to the Commons. Its headline measure cut VAT on theme parks, zoos, and fairs from 20% to 5% between 25 June and 1 September, at a cost of around £300m, according to gov.uk.
The Treasury calls the scheme temporary. Temporary or not, the £300m does not come from nowhere: it is paid by taxpayers, either through tax now or borrowing later.
The Pope's demands now sit with lawmakers. Anthropic's public filing is expected within months, and no British minister has yet published a formal response to the encyclical.
