Pope Leo XIV urges global AI rules, bans lethal machines

Pope Leo XIV on May 25 released a nearly 43,000‑word encyclical titled Magnifica Humanitas (Magnificent Humanity) calling for binding international regulation of artificial intelligence and a prohibition on machines making lethal or otherwise irreversible decisions.

The document warns that the largest AI developers are private, often transnational firms whose resources can exceed those of many governments. The encyclical argues that concentrated technical and economic power can evade public accountability, and it urges nations to adopt clear legal frameworks and independent oversight for AI systems rather than rely on voluntary industry pledges.

The text identifies major risks from AI, including the use of automated systems to spread disinformation, the displacement of workers through automation, and the deployment of autonomous weapons. On artificial systems used in combat, the pope wrote: “It is not permissible to entrust lethal or otherwise irreversible decisions to artificial systems.” The encyclical also warns that AI-driven disinformation could erode democratic processes.

On work and employment, the document says automation is changing the structure of jobs in ways that do not automatically benefit workers. It states that the pursuit of greater profits cannot justify choices that systematically eliminate employment. The encyclical frames a slower, more deliberate approach to AI adoption as an act of responsible care and does not oppose AI development outright.

Christopher Olah, co-founder of Anthropic, appeared at Vatican events around the encyclical’s release as a lay presenter. Olah leads research on interpretability, studying how large language models form internal representations and arrive at decisions, work the document links to demands for transparent, accountable systems subject to meaningful human oversight.

Anthropic has taken visible positions this year: the company contested U.S. defense restrictions in court, promoted an approach to U.S.-China AI policy that keeps safety measures, and published research showing autonomous AI agents can exploit technical vulnerabilities, including in cryptocurrency systems, without human instruction.

Magnifica Humanitas calls on political leaders, technologists, and civil society to cooperate on international agreements and enforcement mechanisms. The encyclical asks that AI development proceed in ways that protect human dignity, preserve democratic institutions, and safeguard livelihoods.