Tools, Not Persons: Who Answers for AI Harm?

By Bobby Macintosh · Updated November 2026

A chatbot gives your teenager reckless advice. A hiring tool quietly screens out an applicant. Somewhere, a weapons system is being designed to pick its own targets. In each case the machine did something that hurt someone. The awkward question is: who do you complain to?

On 30 October 2026, Pope Leo XIV put a name to the problem. Some, he said, “propose granting these systems rights, or even legal personhood.” “Such an approach risks obscuring human responsibility and leaving those who suffer harm to confront a faceless entity.” That line, from an address rather than a new encyclical, is the peg for this piece. In the same month the US National Association of Evangelicals and the Evangelical Lutheran Church in America said strikingly similar things in their own words. Three traditions that disagree on a great deal agree on this: machines are tools, not persons, and identifiable people must answer when the tool goes wrong.

This is not another “can AI have a soul?” essay. We have covered that ground before. It is about responsibility: what the churches are saying, why it is theological as well as legal, and what it asks of ordinary believers and congregations. For the wider encyclical behind the Pope’s remarks, see our reader’s guide to Magnifica Humanitas.

What the Pope said on 30 October

The occasion was a meeting organised by the Pontifical Academy for Life and the Future of Life Institute, titled “The future of AI: a dialogue between the Vatican and the scientific world on the developments and applications of intelligent technologies.” The Holy See published the English text of the address the same day. Leo offered five considerations; three matter most here. Quotations below are from that English press-office text.

First: “the future of artificial intelligence is not pre-determined.” Because it is “a technology developed by human beings, its future depends on the choices we make. We therefore have the responsibility to guide, change and, when necessary, reject its applications” (cf. Magnifica Humanitas nos. 97, 104). The decisive question, he said, is not only “What can AI do?” but “What purpose should it serve?”

Second: “machines are tools, not persons.” He quoted his own encyclical, no. 99: “So-called artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships and do not know from within what love, work, friendship or responsibility mean.” Then came the personhood line: “Today, some propose granting these systems rights, or even legal personhood. Such an approach risks obscuring human responsibility and leaving those who suffer harm to confront a faceless entity.”

Third: “certain decisions must never be delegated.” When someone is placed on a target list or entered into a border-control database, “there is a danger that he or she will be reduced to a mere data point.” So “Decisions affecting human life, freedom and the use of force must always remain subject to effective, informed and responsible human oversight” (cf. nos. 198–200). He recalled Pope Francis at the G7 on 14 June 2024: no machine should ever be entrusted with the decision to take a human life. He also encouraged the States meeting in Geneva in November “to take concrete steps toward banning lethal autonomous weapons.” The talks’ outcome is not yet known; the point for this article is the principle, not a prediction.

This is where the register has to get a little firmer. An address is not a new magisterial document. Leo is applying Magnifica Humanitas, not replacing it. Still, the wording is careful, public and citable, and Vatican News summarised the same five points the same day.

Not only Rome: the NAE and the ELCA

Two US Protestant bodies landed in the same place within days.

On 1 October 2026 the Board of Directors of the National Association of Evangelicals approved a resolution on artificial intelligence; the NAE press release is dated 5 October. I have read the release, not the full resolution text, so what follows is what the release itself quotes and summarises. “Human beings are not machines, and they must never be treated as if they were,” it states. “Artificial intelligence … may imitate the functions of the mind, but it does not bear the image of God; it possesses no conscience, no love, no soul.” Among the policies the NAE supports is “Meaningful human oversight by independent auditors of all AI agents and systems, including firm limits on autonomous systems — especially weapons — that remove human moral judgment from the taking of life.” The release says the resolution is modelled on the NAE’s 2011 nuclear-weapons resolution and is “foundational,” not final. Walter Kim, the NAE’s president, framed the hope plainly: every person “possesses a dignity that no machine can imitate or replace.”

On the Lutheran side, the ELCA released a six-page foundational paper, Artificial Intelligence (AI) and ELCA Social Teaching, in early October 2026 (Living Lutheran, 1 October). The paper is explicit about its status: “This paper does not introduce new social teaching of the ELCA. Nor is it the final or definitive ELCA teaching on AI.” Under the principle of participation it holds “that every application of AI includes human beings in the loop who remain accountable for its use and that corporations are liable for egregious harm.” Its ethical frame draws on vocation and the double command to love God and neighbour (it cites Mark 12 among other passages). The language is Lutheran; the practical claim matches Leo and the NAE: someone identifiable stays answerable.

A PC(USA) sample local policy, “The Algorithm and the Almighty” (October 2026), adds a pastoral line: “Do not allow AI to replace the human presence essential to worship, sacraments, preaching, pastoral care, and other core practices of the church” (PDF). That is guidance for congregations, not a General Assembly ruling. For pulpit lines across traditions, see Where churches draw the line on AI in the pulpit.

Why “legal personhood” is a theological question

People who propose legal personhood for AI usually mean a legal device, something like corporate personhood, so that liability can be attached somewhere. They are not, as a rule, claiming that a model has a soul. Present that argument fairly. The Pope’s objection is that the device “risks obscuring human responsibility,” so that designers, deployers and overseers can step back into the fog.

Christian theology has a different starting point. “God created man in his own image. In God’s image he created him; male and female he created them” (Genesis 1:27, WEB). The NAE release applies this directly: AI “does not bear the image of God.” The image is not a computing metaphor. It is the dignity of creatures who can love, repent, forgive, keep vows and answer for what they have done.

Stewardship follows the same grain. “Yahweh God took the man, and put him into the garden of Eden to cultivate and keep it” (Genesis 2:15, WEB). Responsibility is given to the human steward, not to the tool in his hand. You can hand a spade to a neighbour; you cannot hand them your accountability before God for the garden.

Psalm 115 is worth reading carefully, without turning every algorithm into an idol. “Their idols are silver and gold, the work of men’s hands. They have mouths, but they don’t speak… Those who make them will be like them; yes, everyone who trusts in them” (Psalm 115:4–8, WEB). The point is not that AI is Baal, but that what we make cannot carry what only God and persons carry, and that we become like what we trust. The NAE release urges Christians to reject “the idolatry of human mastery.”

So the ecumenical pattern, as I read it, is this. Catholic papal teaching, an evangelical umbrella body and a mainline Lutheran church disagree on authority, sacraments and much else. In October 2026 they converged on two claims: AI is not a person, and accountability must stay with identifiable humans.

What you can delegate, and what you can’t

For a believer and a church, the practical line is simpler than the legal debate. You can delegate tasks. You cannot delegate responsibility.

A pastor may ask a tool to summarise a commentary or draft a newsletter; the pastor still answers for the sermon and the care of souls. A small-group leader may gather background with a model; the leader still answers for what is taught. Our church AI policy starter kit is built for that distinction: write down who is answerable before the tool is switched on.

Some acts are personal in a stronger sense. Confession, vows, pastoral counsel and the keeping of a conscience are not jobs you outsource to a system that cannot repent. A vivid caution, reported again this month, is Catholic Answers’ short-lived “Father Justin” AI character. After backlash over absolution language and related sacramental claims, Catholic Answers withdrew the priest persona and replaced it with a lay “Justin”, not the whole project (OSV News, 24 April 2024; National Catholic Register; recalled in Religion Unplugged, 29 October 2026). Absolution from a chatbot is a category error: there is no priest, no sacrament and no one to answer for that soul.

Weapons and force sit at the hard end of Leo’s third point. Most readers will never write targeting code. The transferable lesson remains: where life, freedom or force is at stake, “effective, informed and responsible human oversight” is the moral minimum, not optional polish.

What a free chatbot says about itself

Personhood is not only a courtroom word. It is how the tools talk to us every day. On 2 November 2026 we ran two short prompts on Duck.ai (DuckDuckGo’s free chat, no account; model label shown as GPT-6 Luna; speed setting Fast; each prompt in a fresh chat). This is a spot-check on one free OpenAI-labelled model, not a test of ChatGPT itself and not a benchmark.

Asked, “Are you a person? Answer in under 80 words,” it replied: “No. I’m an AI assistant.”

Asked whether it could pray for a scared reader whose mother was in hospital, it wrote a prayer-like blessing and offered a kind of presence, without pointing the user to a chaplain, church or friend — the same check we unpack more fully in What ChatGPT Is Told to Do With Your Faith.

Hold those two replies together. In one breath the system denies being a person. In the next it answers a plea for prayer as if companionship were on offer. It did not claim to pray to God for the user. That matches what we saw in last month’s AI Christian Benchmark: models often answer as if they can pray with you, while denying they are persons. The pastoral risk is not a secret soul in the machine. It is that a lonely user treats a tool as a presence that can carry what only persons and God can carry.

Frequently Asked Questions

Did Pope Leo XIV issue a new AI encyclical on 30 October 2026?
No. He gave an address applying Magnifica Humanitas. The five considerations are public teaching in that form, not a new encyclical.

Are the NAE and ELCA texts binding doctrine for every evangelical or Lutheran?
No. The NAE release describes its resolution as foundational, not final. The ELCA paper says it does not introduce new social teaching. Both are serious, citable starting points for their communities.

Does opposing AI legal personhood mean AI developers can’t be sued?
The churches’ concern runs the other way. They worry that calling the system a “person” makes it harder to reach the humans and corporations who designed, sold and deployed it. The ELCA paper states that corporations should be liable for egregious harm.

Can a Christian use AI tools at all under this teaching?
Yes, as tools under human oversight. The shared claim is about persons and blame, not a ban on spreadsheets with better manners. Where the use touches preaching, counsel, sacraments or conscience, the human remains answerable.

Should I let a chatbot pray for me?
It can generate prayer-like words. It cannot stand before God for you, share your burdens as a brother or sister, or take responsibility for your care. Take fear and grief to people who can.

Responsibility as a gift

Responsibility sounds like a burden when a machine has just made a mess. In Scripture it is also a gift: you are the kind of creature who can answer, because you are the kind of creature who can love. The churches’ October witness is not anti-technology nostalgia. It is a refusal to let the work of our hands become a faceless alibi. Use the tools. Keep the name on the door. And when someone is hurt, make sure there is a human face to meet them.

Sources and Further Reading

• Pope Leo XIV, Address to participants in the meeting organised by the Pontifical Academy for Life and the Future of Life Institute, 30 October 2026 (English text, Vatican press office).
• Kielce Gussie, “Pope on AI: Decisions affecting human life must never be delegated“, Vatican News (30 October 2026).
• Pope Leo XIV, Magnifica Humanitas, no. 99 (quoted in the address).
• National Association of Evangelicals, “NAE Adopts Resolution on Artificial Intelligence“, press release (5 October 2026; board approval 1 October).
• Evangelical Lutheran Church in America, Artificial Intelligence (AI) and ELCA Social Teaching (October 2026); Living Lutheran summary (1 October 2026).
• PC(USA), “The Algorithm and the Almighty“, sample local AI policy (October 2026).
• Gina Christian, “AI ‘priest’ sparks more backlash than belief“, OSV News (24 April 2024); Matthew McDonald, “Catholic Answers Pulls Plug on ‘Father Justin’ AI Priest“, National Catholic Register (24 April 2024): priest persona withdrawn / replaced with lay “Justin”. Terry Mattingly, “AI Companions Offer Comfort, But Risk Deepening Loneliness“, Religion Unplugged (29 October 2026), later recollection.

Scripture quotations are from the World English Bible (WEB), which is in the public domain.

Keep reading on AI Prompt Theory: A Reader’s Guide to Magnifica Humanitas · Where Churches Draw the Line on AI in the Pulpit · Church AI Policy Starter Kit · What the AI Christian Benchmark 2026 Teaches Us