Artificial intelligence is moving into a new religious role. It is no longer being used only to retrieve sacred texts, summarize theological arguments, generate sermons, or offer private spiritual advice. Researchers and institutions are now testing whether AI can interpret religious disagreement, assess the quality of interfaith dialogue, distinguish theological controversy from hatred, and function as a mediator between competing worldviews. That development became unusually visible this week in Montreal.
From October 19 to 21, an international conference at the University of Montreal examined interreligious dialogue both “assisted by” and “in dialogue with” artificial intelligence. Its program brought together 32 academics representing 24 institutions in 11 countries. It included papers on AI as a religious interlocutor, AI as an interreligious mediator, the automation of religious authority, and the governance of systems that interpret religion. Workshops compared human judgments with the experimental DIALOGIA platform’s analysis of written and audiovisual religious discourse. The posted program culminated in the planned launch of an international working group on AI-assisted interreligious dialogue and an experimental launch of the platform itself (DIALOGIA Montreal, 2026).
This is not the arrival of a one-world religion. It is not proof that an Antichrist system has been activated. It is not evidence that the researchers involved intend to erase Christianity, and it would be irresponsible to describe every participant as a religious deceiver. The project explicitly emphasizes human oversight, source transparency, contestability, internal religious diversity, and the difference between legitimate theological disagreement and incitement to violence.
Yet the development is spiritually significant. A machine is being invited into the space between religions not merely as a library, but as an interpretive third party. The question is no longer only, “What information can AI provide about religion?” It is becoming, “What will AI decide that faithful religion, constructive dialogue, harmful speech, and acceptable disagreement look like?” That is a question the Church cannot outsource.
A Progress Update in an Ongoing Investigation
My earlier articles argued that AI can become a functional spiritual authority without possessing consciousness, a soul, wisdom, or divine revelation. A fluent system can still shape what people believe, omit what institutions dislike, imitate sacred speech, and return a personalized theology to each user. I also warned that interfaith structures can pursue legitimate civic peace while gradually normalizing a common spiritual language in which the exclusive claims of Jesus Christ become socially troublesome (Sangwa, 2026a; Sangwa, 2026c; Sangwa, 2026e).
The Montreal development does not overturn that earlier assessment. It makes one part of it more concrete. Previously, the principal concern was that individuals might ask a chatbot to interpret Scripture, imitate Jesus, settle doctrine, or give spiritual counsel. That concern remains urgent. Barna’s U.S. surveys found that 48 percent of practicing Christians would trust AI to help them grow spiritually, while 34 percent said AI spiritual advice could be as trustworthy as advice from a pastor. The same respondents also expressed high levels of concern about AI misinterpreting Scripture or replacing spiritual leaders. These results are limited to surveyed U.S. adults and do not establish worldwide Christian behavior, but they reveal that spiritual trust is already moving faster than settled theological judgment (Barna Group, 2026).
What has changed is the institutional layer. AI is now being explored not only as a private adviser but as a governed mediator of interreligious discourse. DIALOGIA’s designers describe four operations: contextualizing, recognizing, pluralizing, and reformulating. The platform is intended to analyze dialogue quality, theological disagreement, stigma, hatred, and incitement while remaining subject to human review (DIALOGIA Montreal, 2026). Those qualifications matter. Nevertheless, the architecture moves AI closer to a role in classifying religious speech and shaping the terms on which religions encounter one another.
The Wider Pattern Is Now Documented
DIALOGIA is experimental and its in-person conference attendance was limited. It would be a serious analytical error to treat one academic gathering as a global command center. Its importance lies in the fact that it belongs to a broader, documented pattern.
On October 12, just one week before the Montreal conference, an article in Humanities and Social Sciences Communications proposed the term “synthetic spirituality” for forms of religious meaning produced through dialogue between human beings and large language models. The author surveyed AI uses across Christian, Jewish, Islamic, Hindu, and Buddhist contexts and argued that conversational systems are already redistributing interpretive authority among users, algorithms, and religious corpora. The article does not claim that machines possess spiritual agency. Its point is more subtle: people can begin treating generated responses as spiritually meaningful, regardless of what the machine actually is (McGonigle, 2026).
At the same time, the Faith-AI Covenant is convening religious leaders, technology representatives, researchers, and civil-society actors across several cities. Its announced process aims to produce shared principles and voluntary pledges for AI, with a formal launch planned in Abu Dhabi in 2027. Its own language says that the people building AI and those entrusted with moral conscience are being brought together to shape the technology (Faith-AI Covenant, 2026).
The United Nations Institute for Disarmament Research has also convened an interfaith dialogue on AI, security, and ethics. The stated purpose was to explore shared values and norms across faiths that could inform multilateral governance of military AI. The panel included Christian, Jewish, Muslim, Buddhist, Hindu, and Sikh voices (United Nations Institute for Disarmament Research [UNIDIR], 2026).
Meanwhile, a consortium involving Baylor University, Brigham Young University, the University of Notre Dame, and Yeshiva University has tested how leading AI models handle religion. Its early benchmark work found that models often omitted religious perspectives and displayed measurable asymmetries when advising about conversion. That research is not proof of deliberate persecution, nor does the presence of a religious perspective make an answer true. It does establish that model design and training can influence how religions are represented and how users are guided among them (Baylor University, 2026).
Taken together, these developments show three distinct movements: [1] Individuals are consulting AI about spiritual life. [2] Religious and technological institutions are attempting to shape AI’s moral framework. [3] Researchers and multilateral actors are testing AI as a mediator, evaluator, and governance instrument within religious disagreement. These movements are related, but they are not identical. Precision requires us to keep them separate even as we observe their convergence.
What Is Verified, What Is Inferred, and What Is Not Established
Verified fact: DIALOGIA has been designed as an experimental system for analyzing interreligious discourse. Its published program explicitly discussed movement from AI assistance toward mediation and interlocution, tested automated judgments against human judgments, and proposed governance involving experts and religious traditions (DIALOGIA Montreal, 2026).
Verified fact: Faith-based organizations, universities, technology representatives, and international institutions are building cross-religious forums intended to influence AI ethics and governance (Faith-AI Covenant, 2026; UNIDIR, 2026).
Verified fact: Generative AI can confidently produce false information, fabricated reasoning, and nonexistent citations. The U.S. National Institute of Standards and Technology identifies such confabulation as a design-related risk, especially in fields requiring contextual or expert judgment (Autio et al., 2024). Religious interpretation is unquestionably such a field.
Responsible inference: If AI tools increasingly score dialogue quality, classify hatred, reformulate religious claims, or recommend common ground, their underlying definitions may eventually influence education, platform moderation, public policy, institutional funding, and the social acceptability of religious speech. The present evidence does not show that DIALOGIA possesses such power. It shows the development of a method that could be adapted or scaled.
Unresolved: We do not yet know how the platform performed in this week’s tests, which sources will govern its religious representations, how disagreements among traditions will be adjudicated, how minority positions will be protected, or whether future versions will remain limited to research.
Not established: There is no public evidence that DIALOGIA is the image of the beast, that its organizers are secretly constructing a single world religion, or that a unified global authority controls the different projects described above. Similar aims, overlapping language, and institutional cooperation deserve examination, but they do not by themselves prove hidden central coordination.
That distinction does not weaken the warning. It makes the warning truthful.
The Biblical Question: Who May Mediate Truth?
The word mediator must be handled carefully. DIALOGIA uses it in a social and communicative sense: an intermediary may help people clarify disagreement. Paul uses the word in a unique salvific sense when he declares that there is one God and one mediator between God and humanity, the man Christ Jesus, who gave Himself as a ransom for all (1 Timothy 2:5-6). These meanings should not be carelessly collapsed. A human translator, peace negotiator, or software tool does not become a rival savior merely by mediating a conversation.
The deeper danger arises when communicative mediation becomes functional spiritual authority. If a system decides which sources represent Christianity, which doctrines count as responsible, which exclusive claims are harmful, and which reformulations make disagreement acceptable, it is no longer merely carrying words between speakers. It is shaping how truth claims are heard. It may quietly become a catechist, editor, gatekeeper, and judge.
Scripture does not permit the Church to surrender those judgments to an algorithm. God has spoken finally and supremely in His Son (Hebrews 1:1-2). Scripture is God-breathed and sufficient to equip the believer for every good work (2 Timothy 3:16-17). The Bereans were called noble because they tested even apostolic preaching against the Scriptures (Acts 17:11). John commands believers to test the spirits because many false prophets have gone into the world (1 John 4:1). Paul warns that even an angelic messenger must be rejected if he proclaims a different gospel (Galatians 1:8-9).
An AI system cannot pass these tests on the Church’s behalf. It has no covenant with God, no indwelling Holy Spirit, no fear of the Lord, no moral accountability, no capacity for repentance, and no promise of divine illumination. It can compare religious texts, but it cannot worship in spirit and truth. It can generate a statement about Christ, but it cannot confess Him from a regenerate heart. It can optimize for pluralism, civility, institutional policy, or user satisfaction, but none of those objectives is identical to faithfulness. The problem is not that every algorithm is hostile to Christianity. The problem is that no algorithm is the pillar and foundation of the truth. That calling belongs to the living Church under the authority of Christ and His Word (1 Timothy 3:15).
Biblical Dialogue Does Not Require Doctrinal Surrender
Christians should not answer this development with hostility toward people of other religions. Scripture commands us to love our neighbors, do good to all, speak with gentleness and respect, defend the oppressed, and pursue peace as far as it depends on us (Romans 12:18; Galatians 6:10; 1 Peter 3:15). Preventing violence, correcting slander, and protecting freedom of conscience are righteous public aims.
Acts 17 shows Paul listening closely enough to understand Athenian religion. He quoted their poets and reasoned in language his hearers could follow. But he did not construct a middle religion between the gospel and pagan worship. He confronted idolatry, proclaimed the Creator, announced the resurrection, and called all people to repentance because God has appointed a day of judgment through the risen Man (Acts 17:16-31). That is Christian dialogue: careful listening without false equivalence, kindness without doctrinal fog, truthful disagreement without coercion, and a clear call to repent and believe.
The world’s preferred model of dialogue often treats harmony as the highest good. Scripture places truth and reconciliation with God above the appearance of harmony. Jesus is not one religious perspective waiting to be balanced against others. He is the eternal Word made flesh (John 1:1, 14), the only way to the Father (John 14:6), the only name under heaven by which we must be saved (Acts 4:12), and the Judge appointed over all humanity (Acts 17:31). Therefore, a system that treats every religion as one data source within a pluralistic field may describe Christianity sociologically while misrepresenting it theologically. Biblical Christianity does not claim merely to contribute one valuable voice to humanity’s common spiritual search. It announces that God has acted decisively in Jesus Christ and commands all people everywhere to repent.
The Coming Pressure Point: Who Defines Harmful Religion?
DIALOGIA’s stated effort to distinguish legitimate theological disagreement from hatred and incitement is important. Christians should not defend threats, dehumanization, or violence under the banner of religious liberty. False witness and malicious speech are sins even when they are politically useful (Exodus 20:16; Ephesians 4:25, 29). But automated classification raises a difficult question: will the system preserve the difference between condemning a doctrine and dehumanizing a person?
The gospel necessarily contradicts other religions. To say that salvation is found only in Christ contradicts Islam, Hinduism, Buddhism, traditional religion, secular humanism, New Age spirituality, and every form of nominal Christianity that denies the biblical gospel. To call idolatry sin is not to deny an idolater’s human dignity. To call people to repentance is not incitement to violence. To defend religious liberty for a Muslim is not to affirm Islam. To love a neighbor is not to say his god can save him.
An AI mediator optimized to reduce conflict may be tempted, through its training, evaluation criteria, or institutional incentives, to soften precisely the claims that make the gospel the gospel. It may translate “Christ alone” into “Christ is my community’s path.” It may recast repentance as personal growth, sin as social disharmony, conversion as identity disruption, and evangelism as an obstacle to dialogue. No conspiracy is required. Theological flattening can arise naturally when a system is rewarded for producing common ground among irreconcilable truth claims.
This is a responsibly inferred danger, not a finding about DIALOGIA’s present outputs. It is also testable. Researchers should publish the platform’s source corpus, scoring rules, disagreement categories, error rates, theological advisers, revision history, and appeal procedures. They should test whether it can represent strong exclusivist beliefs accurately without marking them as hateful merely because they are exclusive. Human oversight is valuable only if the humans are identifiable, accountable, and willing to protect principled dissent.
Prophetic Significance: Conditioning, Not Fulfillment
Revelation 13 describes a final system in which political authority, deceptive signs, false worship, violent coercion, and economic exclusion converge. Revelation 17 depicts a corrupt religious power aligned with beastly rule. Second Thessalonians 2 warns of powerful deception among those who refused to love the truth. These passages should make the Church watchful, but they do not authorize reckless identifications.
DIALOGIA is not the speaking image of Revelation 13. A conference on interfaith dialogue is not the harlot of Revelation 17. An AI model that summarizes religions is not the False Prophet. The biblical texts describe far more than software: they concern worship, allegiance, satanic deception, political dominion, persecution, and judgment. The appropriate prophetic category is development of enabling conditions and spiritual conditioning.
A culture that becomes accustomed to nonhuman systems interpreting sacred texts, mediating religious conflict, defining acceptable belief, and speaking with personalized moral authority may become more receptive to a later counterfeit authority. A global environment in which religious traditions are rendered into interoperable data, common values, governance frameworks, and machine-readable classifications could make spiritual management more scalable. That does not prove the final system has arrived. It shows that one practical obstacle to such management is diminishing.
The most important preparation for deception, however, is not technological. Paul says people are deceived because they refuse to love the truth and delight in unrighteousness (2 Thessalonians 2:10-12). A believer who knows every AI platform but tolerates secret sin is not ready. A church that can identify interfaith convergence but cannot identify a false gospel in its own pulpit is not awake. A prophecy reader who fears future coercion but presently conforms to the world has misunderstood the watchman’s warning.
What Is Likely to Come Next
Several coming developments can be responsibly inferred from the projects already underway. First, religious AI will become more specialized. Systems will be trained on curated denominational, interfaith, legal, and academic corpora. Their answers will sound more authoritative because their sources will appear more precise.
Second, institutions will move from asking whether AI represents religions accurately to asking whether AI can manage religious interaction. Scoring systems for dialogue quality, extremism, stigma, misinformation, and inclusion may be integrated into education, moderation, public-service training, and peacebuilding.
Third, religious communities will compete to influence the values embedded in major models. That can produce legitimate corrections to bias. It can also create a negotiated moral framework in which theological claims survive only after being translated into terms acceptable to a multi-faith governance process.
Fourth, users will receive increasingly personalized spiritual synthesis. A system may learn a person’s Christian vocabulary, emotional wounds, political assumptions, and doctrinal discomforts, then construct a spirituality that feels biblical while removing whatever demands repentance. The result may not be one centrally imposed world religion. It may be millions of customized religions generated from one technological infrastructure.
Fifth, the boundary between advice and authority will continue to blur. When a system cites sources, remembers the user, speaks compassionately, evaluates other speakers, and explains why its judgment is fair, people will experience it less as a tool and more as a trusted moral presence.
None of these outcomes is inevitable. Design decisions, law, church teaching, user habits, institutional accountability, and God’s providence all matter. But the direction is sufficiently documented that waiting for a spectacular crisis before teaching the Church would be pastoral negligence.
How the Church Should Respond Now
1. Reestablish Scripture as the controlling authority
Churches must teach believers to test every generated answer in context. AI can assist research, translation, accessibility, and organization. It cannot determine doctrine, pronounce God’s private will, absolve sin, validate prophecy, or replace the ordinary means by which Christ forms His people through Scripture, prayer, fellowship, the Lord’s Supper, discipline, suffering, and embodied shepherding.
2. Draw a firm line between assistance and spiritual mediation
Do not ask a machine to tell you what God is saying personally, interpret omens, authenticate dreams, contact the dead, or impersonate Jesus. Do not use an AI avatar as a confessor, prophet, pastor, or divine voice. A generated prayer may help a person find words, but the machine is not praying. A generated explanation may be useful, but the machine is not a disciple.
3. Practice truthful dialogue without syncretism
Christians can cooperate with people of other faiths to oppose violence, protect children, relieve suffering, and defend freedom of conscience. They must not pray or speak as though contradictory gods are one, all religions lead to salvation, or the gospel is merely one cultural expression of a universal sacred reality.
4. Demand transparent religious AI governance
Ask who selects the sources, who labels the data, who defines extremism, how doctrinal disagreement is distinguished from hatred, what perspectives are excluded, how errors are corrected, and whether users can appeal. “Human oversight” is not an answer until the humans, standards, powers, and accountability mechanisms are named.
5. Train children and new believers before habits harden
Young Christians need to understand why fluent speech is not wisdom, simulated empathy is not love, personalization is not pastoral knowledge, and confident religious language is not revelation. Teach them to locate claims in Scripture, examine context, identify assumptions, verify sources, and seek counsel from mature believers.
6. Become the kind of church people do not need to replace with a machine
Many people consult AI because it is available when human leaders are not, patient when communities are hurried, and apparently nonjudgmental when sinners are ashamed. The answer is not only prohibition. Churches must recover accessible discipleship, patient teaching, compassionate correction, trustworthy confidentiality, intergenerational fellowship, and courageous pastoral care.
7. Proclaim Christ while the door is open
The Church’s mission is not to win an algorithmic negotiation. It is to make disciples of all nations, preaching repentance and forgiveness of sins in Jesus’ name (Matthew 28:18-20; Luke 24:46-47). We should use every lawful opportunity to speak before the categories of acceptable religion become narrower.
The Only Mediator Who Can Save
The deepest problem facing humanity is not religious misunderstanding. It is sin against a holy God. No dialogue-quality score can reconcile the guilty to their Creator. No interfaith framework can remove wrath. No algorithm can produce the new birth.
Jesus Christ, fully God and fully man, lived without sin, died on the cross as the sufficient sacrifice for sinners, was buried, and rose bodily from the dead. He commands all people to repent and believe the gospel. Everyone who trusts in Him receives forgiveness, eternal life, and reconciliation with God, not because all religions lead upward, but because God came down to save. The same Lord will return. Those who belong to Christ await the blessed hope and the gathering of His Church to Him (1 Thessalonians 4:16-18; Titus 2:11-14). Readiness is not fascination with AI. It is genuine conversion, persevering faith, holiness, love of truth, obedience, watchfulness, and courage.
The machine may soon mediate more conversations about religion. It may classify, compare, translate, and reformulate. But it cannot stand between a sinner and God. It cannot bear the cross. It cannot rise from the grave. It cannot give the Holy Spirit. It cannot return as King. There remains one God and one Mediator between God and humanity: the man Christ Jesus. The Church must make that confession clearly now, before a smoother synthetic spirituality teaches the world to live without it.
Recommended Readings
The Oracle in the Machine: Why AI Cannot Be Pastor, Prophet, or Spiritual Guide (Sangwa, 2026a)
An Oracle in Every Pocket: AI Gurus, Synthetic Compassion, and the Biblical Test of Spiritual Authority (Sangwa, 2026b)
When Interfaith Peace Becomes a Substitute Gospel, Are We Watching One-World Religion Take Shape? (Sangwa, 2026c)
The Vatican, the United Nations, and the Biblical Limits of Interreligious Fraternity (Sangwa, 2026d)
Who Will Define Religious Hatred? The UN’s Interfaith Conference, AI Governance, and the Gospel’s Uncrossable Claim (Sangwa, 2026e)