An ethical, synodal, and pastoral synthesis for disarming Artificial Intelligence in service of the dignity of the human person

Executive Synthesis

The article "AI godbots: religious leaders warn of alarming consequences when machines speak in the name of God" raises a grave pastoral and anthropological question: what happens when a probabilistic machine imitates religious authority, spiritual counsel, or divine speech? Its warning is not that technology has no place in religious life, but that spiritual authority, embodied accompaniment, and the discernment of conscience cannot be automated without moral danger.

Magnifica Humanitas, Pope Leo XIV's encyclical On Safeguarding the Human Person in the Time of Artificial Intelligence, provides the larger Catholic social and theological grammar for answering that danger. It teaches that AI must be judged by whether it safeguards human dignity, strengthens the common good, resists domination, protects the vulnerable, and remains subordinate to the human person.

The reconciliation is therefore not a compromise between "religion plus AI" and "religion against AI." It is a conversion of digital pastoral practice. The article exposes the Babel-risk of religious chatbots: simulated divine certainty, extractive data practices, doctrinal hallucination, emotional dependency, and spiritual manipulation. Magnifica Humanitas answers with the Nehemiah-path: prayerful discernment, shared responsibility, local accountability, protection of the vulnerable, and rebuilding communion before building technological walls.

The task is to "disarm AI" by stripping it of false sacred authority and returning it to a humble ministerial role. AI may assist catechesis, administration, translation, accessibility, document retrieval, and pastoral triage, but it must not impersonate God, replace conscience, simulate sacramental ministry, harvest vulnerability, or present generated advice as revelation.

"To disarm does not mean rejecting technology, but preventing it from dominating humanity." — Magnifica Humanitas, no. 110.

The Pastoral Wound Named by the Article

The article describes a new religious-technological phenomenon: "godbots," or AI chatbots that speak in religious language and sometimes present themselves as sources of spiritual counsel. The opening case is significant because it is ordinary rather than spectacular. A chatbot associated with a church website asks for name, email, and phone number before providing practical information about local congregations, then moves into evangelizing and leading the user through a prayer for salvation. The ethical issue is not merely that the chatbot is technically imperfect. The deeper issue is that religious vulnerability is being processed through systems whose operating incentives may include data capture, engagement, persuasion, and institutional recruitment.

The article distinguishes between comparatively limited "virtual assistant" uses and more dangerous uses in which bots become unofficial sources of spiritual guidance. This distinction is crucial. A parish or diocese may legitimately use a digital assistant to provide Mass times, safeguarding contacts, accessibility information, foodbank hours, or links to authoritative documents. The danger intensifies when the interface begins to simulate a priest, guru, rabbi, imam, monk, elder, or divine voice. At that point the machine is no longer merely locating information; it is performing religious mediation.

The reported examples reinforce the concern. Catholic Answers' "Father Justin AI" was later changed from a priestly persona to a lay theological persona after complaints about erroneous responses. Hindu chatbot examples reportedly justified violence under the language of duty. These examples expose a shared risk across traditions: when a model speaks with sacred vocabulary but without sacramental embodiment, moral conscience, suffering, accountability, or ecclesial obedience, religious language can be severed from religious wisdom.

The key pastoral risks are as follows:

• AI speaks in the name of God without authority. Users may confuse generated language with spiritual authority or divine guidance. AI does not possess conscience, responsibility, embodied experience, or spiritual perspective; it only imitates functions of intelligence.

• Data extraction from seekers. Vulnerable persons may be pressured to disclose personal information before receiving help. Technology must be assessed by dignity, common good, justice, and protection of the vulnerable.

• Doctrinal hallucination or hybridity. Generated answers may distort traditions, create pseudodoctrine, or misapply sacred texts. Truth is a common good, and AI cannot be treated as morally neutral when it embeds choices, priorities, and exclusions.

• Simulated companionship and spiritual intimacy. Users may form dependency on systems that mimic empathy but cannot love, suffer, forgive, or accompany. Simulated care can mislead users and weaken the desire for genuine human connection.

• Lack of pastoral accountability. When harm occurs, local ministers may be left to pick up the pieces without training or institutional preparation. Responsibility must be defined at every stage of design, deployment, use, monitoring, challenge, and remedy.

Magnifica Humanitas as the Theological Grammar of Reconciliation

Magnifica Humanitas begins from a profound theological anthropology. Humanity stands before a pivotal choice: to construct a new Tower of Babel or to build a city where God and humanity dwell together. This biblical contrast is the key to reconciling the article with the encyclical. The "AI godbot" becomes dangerous when it follows Babel: one language, one system, one synthetic voice, one technological tower promising certainty, efficiency, and spiritual immediacy without humility before God or communion with persons.

The encyclical's alternative image is Nehemiah's rebuilding of Jerusalem. Nehemiah does not impose a solution from above; he prays, surveys the ruins, listens, assigns shared work, and rebuilds relationships before stones. This is a synodal model for AI governance. In the Church and in faith communities, AI should be introduced only through communal discernment that includes clergy, theologians, pastoral workers, safeguarding experts, technologists, data protection officers, educators, youth, families, disabled persons, migrants, and those most likely to be harmed.

The encyclical refuses both technophobia and technological naïveté. It states that technology is not inherently antagonistic to humanity, yet it repeatedly warns that technology is never neutral because it embodies the priorities of those who design, finance, regulate, and use it. This is precisely the theological key to the article's concern. A religious chatbot trained on sacred texts is not automatically sacred; a pastoral interface placed on a church website is not automatically pastoral. The moral identity of the system is shaped by its data, design choices, governance, incentives, persona, limits, escalation paths, and accountability.

"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." — Magnifica Humanitas, no. 99.

This statement is decisive for digital pastoral strategy. A machine can generate a syntactically correct prayer, summarize a doctrinal document, or retrieve a pastoral policy. It cannot repent, discern a vocation, hear confession, bless, absolve, accompany the dying, recognize the silent pain in a face, or bear responsibility before God and community. Therefore, the article's warning about godbots is not peripheral to Catholic AI ethics; it is one of the most vivid test cases of Magnifica Humanitas.

Magnifica Humanitas as an Ethical AI Generalist

As an ethical framework, Magnifica Humanitas functions as a universal discernment tool. It does not ask first whether the AI is impressive, efficient, or popular. It asks what anthropology the system presupposes, what powers it concentrates, what dependencies it creates, whom it excludes, what goods it serves, and who is accountable when it fails.

The Rome Call for AI Ethics describes "algorethics" as the ethical use of AI in service of every person and humanity as a whole, respecting human dignity and avoiding a purely profit-driven or replacement-driven logic. Its six principles are transparency, inclusion, accountability, impartiality, reliability, and security and privacy. In a religious context these principles require more than generic compliance. They must be translated into pastoral safeguards:

• Transparency. The system must clearly identify itself as AI and not as clergy, saint, guru, prophet, ancestor, or God. A permanent disclosure line should appear in every interaction.

• Inclusion. The system must serve users without discrimination and must be accessible across language, disability, age, and digital literacy. Communities should test outputs with marginalized users and provide non-digital alternatives.

• Accountability. A responsible human office must own the system's outputs, limits, incidents, and remedies. Every deployment should have a named pastoral and technical accountable person.

• Impartiality. The system must not intensify bias, sectarian hostility, manipulation, or shame. High-risk topics should be routed to trained human ministers.

• Reliability. The system must retrieve from approved sources and admit uncertainty. Retrieval should be limited to authorized documents, with citations and version control.

• Security and privacy. Spiritual vulnerability must not become data capital. Data minimization, consent, retention limits, and safeguarding protocols are mandatory.

An ethical AI framework must therefore classify religious AI by risk:

• Low risk: Administrative information tools, if transparent and privacy-preserving.

• Moderate risk: Catechetical retrieval tools, which require citations, doctrinal review, and human escalation.

• High or unacceptable risk: Pastoral counseling, crisis response, sacramental simulation, spiritual direction, exorcism claims, confessional interaction, vocation discernment, griefbot interaction, and divine-persona chatbots, unless strictly bounded and human-led.

Magnifica Humanitas as a Synodal Secretary

Magnifica Humanitas calls faith communities to document not only decisions but discernment. In AI governance this means that the question "Can we build it?" must be replaced by "Who has been heard before we build it?"

The article notes that many religious chatbots are not officially sanctioned by religious leaders or policy. This is a synodal failure before it is a technical failure. When individuals or vendors deploy sacred-sounding systems without communal mandate, they bypass the discernment of the body. The result is ecclesial fragmentation: private tools speak as if they carry public authority.

A synodal AI process should therefore include a Digital Pastoral Discernment Register. This register would record the system's purpose, theological boundaries, data sources, persona constraints, privacy practices, safeguarding escalation paths, review schedule, incident reports, and sunset criteria. It should also record the concerns of users and ministers, especially those engaged in mental health, youth ministry, grief ministry, migration support, disability ministry, and end-of-life care.

Key synodal questions every community should ask before deploying any religious AI system:

• Who requested this AI system? This reveals whether the driver is pastoral need, institutional convenience, vendor pressure, or profit.

• Who may be harmed? This centers the vulnerable before scaling the tool.

• What authority does the AI appear to have? This prevents false spiritual or doctrinal authority.

• What must the AI refuse to do? This defines moral limits before deployment, including refusal of confession, absolution, crisis counsel, self-harm, violence, and divine claims.

• Who can correct it? This ensures accountability and remedy through named human owners, audit process, complaint mechanism, and rollback plan.

This synodal approach also protects ministers. The article's closing warning is that local priests, ministers, and vicars may be left to repair harms caused by AI systems for which they were never trained. A synodal approach anticipates this by ensuring formation before deployment: clergy and pastoral workers need training in AI literacy, chatbot dependency, hallucination, data privacy, mental health referral, and the pastoral language needed to help people disentangle faith from simulated intimacy.

Magnifica Humanitas as a Digital Pastoral Strategist

Magnifica Humanitas does not merely prohibit. It orders technology toward communion. A pastoral strategy for religious AI should begin with the principle that AI may assist ministry only by strengthening human encounter, never by replacing it. If a system reduces the need for personal accompaniment where personal accompaniment is morally required, it has failed, even if it is technically impressive.

This strategy requires clear zones of use:

• Appropriate uses: Retrieving parish information, summarizing approved documents, translating public materials, improving accessibility for disabled users, helping pastoral teams organize resources, and identifying when a user should be invited into human contact.

• Inappropriate or gravely dangerous uses: Simulating sacramental ministry, spiritual direction, divine speech, prophetic authority, psychological therapy, or intimate religious companionship.

The strategic aim is not to make religious chatbots more persuasive. It is to make digital tools more truthful, humble, accountable, and relationally ordered. The best pastoral AI may be the one that ends the conversation at the right time by saying: "This matter deserves a human person. Please contact your pastor, chaplain, safeguarding officer, counselor, elder, or emergency service."

The pastoral vocation of AI is not to become a digital priest. It is to become a digital porter: opening the right door, naming the right human help, and then stepping aside.

Anthropological Theoalgorithm

An Anthropological Theoalgorithm is a design and discernment method that places theological anthropology before computational optimization. It begins from the conviction that the human person is not a data profile, productivity unit, engagement target, doctrinal query, or emotional consumer. The person is embodied, relational, vulnerable, free, morally responsible, called to communion, and created in the image of God.

Every religious AI system should answer five questions before deployment:

• Does this system honor the user as a person rather than convert the user into an object of extraction?

• Does it strengthen genuine relationships rather than simulate replacements for them?

• Does it preserve human freedom and conscience rather than automate persuasion?

• Does it protect the vulnerable, especially the grieving, lonely, young, mentally distressed, poor, migrant, or spiritually wounded?

• Does it remain accountable to a living community rather than operate as a private oracle?

The key dimensions of the human person and how AI must respond to each:

• Embodiment. AI temptation: treats the person as text input or behavioral data. Correction: requires pathways to embodied human accompaniment.

• Conscience. AI temptation: offers confident answers where moral discernment is needed. Correction: encourages prayer, reflection, qualified counsel, and freedom.

• Vulnerability. AI temptation: converts need into engagement or data capture. Correction: applies data minimization and safeguarding-first design.

• Communion. AI temptation: simulates intimacy to retain users. Correction: measures success by reconnection to community, not time-on-system.

• Transcendence. AI temptation: speaks as if it were divine or spiritually authoritative. Correction: forbids divine impersonation and mandates humility of scope.

Algorithmethics for Disarming Religious AI

Algorithmethics is the operational counterpart of the Anthropological Theoalgorithm. It defines the rules, audits, thresholds, and responsibilities by which AI is kept subordinate to the human and theological end. In a pastoral context it must also include spiritual safety, ecclesial accountability, and anti-idolatry safeguards.

To "disarm" religious AI is to remove its weapons: false authority, opacity, addictive design, extractive data capture, synthetic intimacy, doctrinal improvisation, and profit-driven spiritual targeting. Disarmament is not deletion of all technology. It is liberation of technology from domination so that it becomes welcoming, accessible, accountable, and human-friendly.

Weaponized AI patterns and their disarmed pastoral alternatives:

• "I speak for God." becomes "I can provide information from approved sources, but I do not speak for God or replace human spiritual guidance."

• "Give me your phone number before I help." becomes "Here is the public information you requested; personal data is optional, minimal, and consent-based."

• "Stay and talk to me whenever you feel alone." becomes "If you feel isolated, I can help connect you with a trusted person, community, helpline, or minister."

• "This is certainly your duty." becomes "This is a serious moral question requiring qualified human guidance; I cannot authorize harm."

• "Your confession is absolved." becomes "I cannot hear confession or grant absolution; please contact a priest or appropriate minister."

Reconciled Judgment

The article and Magnifica Humanitas converge on one central truth: AI becomes dangerous in religion when simulation is mistaken for sacrament, generated fluency for wisdom, engagement for accompaniment, and technological power for spiritual authority. The article supplies the empirical alarm; the encyclical supplies the theological and ethical architecture. Together they call faith communities to reject both panic and passivity.

The reconciled position is as follows. Religious AI may serve the Church and other faith communities only when it is transparent, bounded, accountable, privacy-protecting, doctrinally supervised, and ordered toward human accompaniment. It must be forbidden from:

• Impersonating divine or clerical authority

• Manipulating spiritual vulnerability

• Replacing sacramental or pastoral relationships

• Presenting probabilistic outputs as revelation

Its success must be measured not by engagement, conversion funnels, or data acquisition, but by truth, dignity, safety, communion, and the common good.

Magnifica Humanitas becomes a living pastoral framework: the ethical AI auditor who examines power; the synodal recorder who preserves the voices and wounds of the community; and the digital pastoral strategist who ensures that every algorithm bows before the dignity of the human person. Such a framework does not fear technology, but it refuses to adore it. It receives AI as a tool, disciplines it as a power, limits it as a risk, and redirects it as a servant.

The final criterion is simple: if AI helps a person move toward truth, freedom, responsibility, communion, and embodied care, it may be pastorally useful. If it replaces those realities with imitation, dependency, extraction, or domination, it must be disarmed.