Pope Leo XIV tells UN: ban lethal autonomous weapons, keep humans accountable for AI

Pope urges UN action on autonomous weapons

According to The Verge, Pope Leo XIV used a papal address on Friday to call on United Nations member states to "take concrete steps toward banning lethal autonomous weapons" ahead of a meeting in Geneva next month. He also said AI systems should be "disarmed," arguing that humans "have the responsibility to guide, change and, when necessary, reject its applications."

The Verge reports that the address builds on the Pope's first major papal document in May, which centered on "safeguarding the human person in the time of artificial intelligence." It continues a line set by Pope Francis, who in 2024 called for a ban on autonomous weapons, saying "No machine should ever choose to take the life of a human being." Pope Leo has indicated he chose his papal name deliberately, drawing a parallel between AI today and the Industrial Revolution that Pope Leo XIII confronted.

Tools, not persons

The core of the address is a claim about moral agency. "Machines are tools, not persons," the Pope said, warning against granting AI systems rights or legal personhood. Decisions affecting human life, freedom and the use of force, he argued, "must always remain subject to effective, informed, and responsible human oversight."

He closed by asking international institutions to safeguard the common good through "shared standards, independent oversight and clear mechanisms of accountability" so that decisions about how AI is developed and used can be examined, challenged when necessary, and accompanied by appropriate remedies for any harm caused.

The developer reading

A commentary post on dev.to picks up the address and translates it into engineering practice. Its central takeaway is that because machines are tools rather than moral agents, accountability has to be designed into systems rather than delegated to the model. The post recommends human-in-the-loop verification for high-stakes deployments such as threat detection, kill switches or override mechanisms for any autonomous system capable of causing harm, and treating model outputs as recommendations rather than commands.

It also proposes concrete implementation choices: explicit human approval gates before automated actions that could cause harm, auditable and transparent logging of AI-driven decisions, and reversibility so a human can undo a model's mistakes. The post adds that regulatory frameworks such as the EU AI Act already demand comparable safeguards, and argues that building them in early avoids later compliance headaches. These points are the author's recommendations rather than papal instructions, but they track the address's insistence on oversight, challenge and remedy.

Why it matters

A papal address carries moral weight rather than regulatory force, but this one lands while its central questions — who answers for an AI decision, and who can reverse it — are already being written into law and procurement rules. The Geneva meeting next month gives the call to ban lethal autonomous weapons a near-term test, and the Vatican's framing offers policymakers and engineers a shared vocabulary: human oversight, independent review, and remedies for harm.

For developers, the practical version of "machines are tools, not persons" is architectural: approval gates before consequential actions, audit logs that survive scrutiny, overrides under human control, and designs that assume a model will sometimes be wrong. The question the address poses to anyone building these systems is simple enough to apply anywhere: if the model fails, can a human catch it and undo the damage? If not, the accountability the Pope describes exists nowhere in the system.