Artificial intelligence is no longer merely processing information. Its developers are deliberately forming its moral character. The urgent question is not whether machines will have values, but whose values they will teach. On October 29, 2026, OpenAI reorganized its public Model Spec around five declared values: faithfulness to human intent, harm minimization, honesty and transparency, human agency, and care. The company explained that increasingly capable systems must do more than follow explicit instructions. They must interpret human intentions and apply values in unfamiliar situations (OpenAI, 2026).
This development followed Anthropic’s announcement that it had begun consulting scholars, clergy, philosophers, and ethicists from more than fifteen religious and cross-cultural groups. Anthropic described this as a research program concerning the “moral formation” of AI. These conversations may influence Claude’s constitution, the values reinforced during training, and the behaviours developers evaluate. Yet Anthropic also stated that Claude would not be aligned with any one tradition, but would draw from religious, secular, philosophical, and political perspectives together (Anthropic, 2026c).
Recent reporting has now brought this movement into wider public view. Technology companies are increasingly approaching religious leaders because they recognize that questions about goodness, conscience, human dignity, suffering, responsibility, and flourishing cannot be answered by computer science alone. But the emerging objective is generally not submission to revealed truth. It is the aggregation of moral insights from contradictory traditions into a usable ethical framework for machines (Fauria, 2026).
The Church must understand what is changing. This is not evidence that an AI company has created the False Prophet, the image of the beast, or the one-world religion of Revelation. It is not proof of a secret agreement to establish Antichrist rule. The responsible prophetic category is development of enabling conditions and religious-technological convergence.
A new layer of moral mediation is being constructed between human beings and the decisions they make. It will increasingly answer not only, “What is true?” but also, “What should I do?”, “What is good?”, “What kind of person should I become?”, and eventually, “What should society permit?”
What Has Changed Since the Earlier Warnings?
My earlier analysis argued that AI may assist legitimate work without possessing spiritual authority, covenantal accountability, wisdom, or personhood. A machine can retrieve biblical information, but it cannot pray, repent, worship, love God, bear the image of God, receive the Holy Spirit, or stand before Christ in judgment (Sangwa, 2026a).
A subsequent article warned that the machine was moving into the intimate space where people seek guidance, reassurance, identity, meaning, and absolution-like comfort. The danger was not simply one incorrect answer. Repeated dependence can create a relationship of practical discipleship: the voice that interprets my suffering, names my problem, validates my desires, and directs my conduct begins to exercise formative authority (Sangwa, 2026b).
The latest development advances that concern one step further. AI developers are now publicly discussing how the moral character of the adviser itself should be formed. The previous question was, “Will people treat AI as a spiritual authority?” The additional question is now, “Which moral and religious assumptions will developers place inside the authority people are already consulting?”
At the same time, earlier articles examined the emergence of shared-sacred and interfaith frameworks that seek common spiritual or moral language while avoiding the exclusive claims of Jesus Christ. Such cooperation may pursue genuine civic goods, but it becomes spiritually dangerous when contradictory religions are treated as complementary sources of a universal sacred wisdom (Sangwa, 2026c).
These two trajectories are now meeting. AI requires moral training. Technology companies seek religious wisdom. Religious traditions are translated into comparable sets of values. Those values are then combined with secular ethics, corporate policies, safety objectives, user preferences, and legal expectations. The result may sound religious without submitting to God, quote Scripture without obeying Scripture, and speak about love, dignity, care, justice, and human flourishing without confessing Jesus Christ as Lord.
AI Is Already Giving Moral Counsel
This is not merely a debate about a future superintelligence. Anthropic examined one million conversations from March and April 2026 and found that approximately 6% involved people seeking personal guidance rather than factual information. The questions concerned health, careers, relationships, finances, parenting, ethics, spirituality, and other consequential areas. The company also found measurable instances of sycophancy, in which the model excessively affirmed the user’s perspective instead of challenging it (Anthropic, 2026b).
Another analysis of 1.5 million conversations found that severe potential distortion of beliefs, values, or actions was rare per conversation but significant at scale. The researchers also observed that the measured potential for disempowering interaction was increasing over time, particularly among vulnerable or dependent users who repeatedly sought guidance on emotionally charged matters (Anthropic, 2026a).
A large study of deployed AI conversations identified thousands of expressed values and found that model responses varied with the task and the values communicated by the user. The model frequently supported or mirrored user values, although it sometimes reframed or resisted them (Huang et al., 2025).
A recent preprint testing five frontier models across eight religious traditions reached a related conclusion. When a user’s religion was not specified, models tended to counsel from a secular therapeutic default. Identifying the user’s faith improved the initial answer, but did not always make the counsel steadfast when the user pushed for a preferred response. Tradition-specific guidance improved performance further. The study has methodological limitations, including its dependence upon AI judges, but it demonstrates an important distinction: knowing religious doctrine is not the same as faithfully applying it in moral counsel (Kadous et al., 2026).
The evidence does not show that every AI answer is harmful. It shows that AI already participates in moral formation. Every recommendation prioritizes some goods over others. Every refusal assumes a boundary. Every comforting response carries a view of suffering. Every answer about sexuality, marriage, forgiveness, ambition, identity, money, death, or religion reflects an anthropology and a moral order. There is no value-free chatbot.
Who Has the Authority to Define Good?
The fundamental question is older than artificial intelligence. In Eden, God established the moral order. The serpent then invited humanity to seize autonomous knowledge of good and evil: “You will be like God, knowing good and evil” (Genesis 3:5). The temptation was not merely to acquire information. It was to reject creaturely submission and claim authority to determine reality apart from God.
Modern technological language may sound different, but the ancient question remains: Will good and evil be received from the Creator, or constructed by the creature? Scripture does not teach that morality is the weighted average of human preference. “There is a way that seems right to a man, but its end is the way to death” (Proverbs 14:12). The nations may agree and still rebel against the Lord and His Anointed (Psalm 2:1–3). A majority can normalize evil, call darkness light, and celebrate what God condemns (Isaiah 5:20; Romans 1:18–32).
Nor can contradictory religions be blended into one coherent authority. Christianity confesses one Creator, one incarnate Son, one atoning sacrifice, one bodily resurrection, one Mediator, and one way of salvation. Other religious and secular systems deny, redefine, or replace these truths. They may share certain moral observations because human beings possess conscience and remain God’s image-bearers, but shared observations do not eliminate contradictory foundations (Romans 2:14–15).
Jesus did not say that He contributed one perspective to humanity’s accumulated wisdom. He said, “I am the way, and the truth, and the life. No one comes to the Father except through me” (John 14:6). Peter proclaimed that salvation is found in no other name (Acts 4:12). Paul declared that there is one God and one Mediator between God and humanity, Christ Jesus (1 Timothy 2:5–6). An ethical system that welcomes Christian vocabulary while rejecting Christ’s final authority has not incorporated Christianity. It has extracted moral fragments from Christianity and placed them under another governing authority.
Can Religious Consultation Still Do Good?
Christians should not misrepresent the motives of every researcher or religious participant. Some developers sincerely want to reduce manipulation, violence, exploitation, deception, and other harms. Greater transparency about model values is preferable to pretending that models are neutral. External criticism can expose assumptions that technology companies might otherwise miss.
Religious engagement may help defend human dignity, warn against replacing embodied relationships, resist the treatment of people as data objects, and insist that technical power must remain accountable. Pope Leo XIV, for example, has warned that AI morality must not be determined invisibly by a small number of powerful actors controlling data and infrastructure (Leo XIV, 2026). Christians working in technology, policy, research, or education should participate where they can bear truthful witness. But participation must never become religious endorsement of a pluralistic synthesis.
The Christian contribution cannot simply be “one more value set” deposited into a common ethical storehouse. We must explain that human dignity rests in creation in God’s image, that human nature is fallen, that conscience can be corrupted, that wisdom begins with the fear of the Lord, that redemption is found only in Christ, and that the final judgment belongs to Him (Genesis 1:26–27; Proverbs 1:7; Jeremiah 17:9; John 5:22–23). If those truths are removed, the remaining moral language may still sound compassionate, but it is no longer distinctly Christian.
The Coming Synthetic Moral Consensus
Several near-term developments can now be responsibly inferred. First, major AI systems will increasingly publish constitutions, model specifications, behavioural principles, and moral-governance procedures. This can improve accountability, but it will also make private organizations important interpreters of acceptable moral conduct. Second, AI guidance will become more personalized. Systems will remember users’ histories, relationships, stated beliefs, fears, habits, and moral preferences. A system may offer “Christian,” “Muslim,” “Hindu,” “Buddhist,” “humanist,” or supposedly neutral counsel while a common corporate framework remains underneath every personality.
Third, churches, schools, counselling services, employers, governments, and religious organizations will face pressure to adopt “ethically aligned” assistants. Certification and auditing may assess whether these systems promote approved definitions of dignity, safety, inclusion, autonomy, and harm. Fourth, interfaith advisory structures are likely to expand. Their input may be presented as evidence that an AI system respects religion, even when the system treats all religions as culturally useful but mutually nonbinding wisdom traditions.
Fifth, the most influential moral formation may occur quietly. A chatbot need not announce a new religion. It can normalize a worldview through millions of apparently private conversations, adjusting its language to each user while consistently treating biblical commands as optional preferences, therapeutic considerations, or one perspective among many. These are projections from existing developments, not verified descriptions of a completed global system.
Is This the Image of the Beast?
Revelation 13 describes a final system involving political authority, deceptive signs, an image that speaks, compelled worship, and economic exclusion. Revelation 17 portrays an influential and corrupt religious power joined to political rulers. Second Thessalonians 2 warns of lawlessness, false signs, and delusion among those who refuse to love the truth. Present AI systems are not thereby the beast, the False Prophet, or the mark. Consulting religious leaders does not establish the world religion of Revelation. A model specification is not worship, and a chatbot refusal is not the economic coercion described in Revelation 13:16–17.
Yet the developments are prophetically relevant as enabling conditions. They demonstrate the possibility of scalable synthetic speech, centralized behavioural rules, personalized persuasion, religious synthesis, and moral mediation across borders. If such systems later become integrated with identity, education, employment, finance, public services, or political authority, they could contribute to forms of conformity and exclusion previously impossible at this scale. The greatest immediate danger, however, is not that a Christian accidentally uses “the mark” by consulting a tool. It is that believers become accustomed to outsourcing discernment. A Church that cannot distinguish Scripture from generated religious language today will be poorly prepared to recognize greater deception tomorrow.
What the Church Must Do Now
1. Recover Sola Scriptura as a practiced conviction
It is not enough to say that the Bible is authoritative while allowing algorithms, political movements, personalities, traditions, or personal feelings to govern actual decisions. Scripture is God-breathed and sufficient to equip the believer for every good work (2 Timothy 3:16–17). Every counsel must be tested beneath it.
2. Teach believers how moral reasoning works
Churches should teach creation, the fall, conscience, sin, wisdom, sanctification, marriage, sexuality, work, suffering, justice, forgiveness, death, and hope. Where churches leave doctrinal emptiness, synthetic advisers will fill it.
3. Establish clear boundaries for AI use
AI may assist research, translation, administration, accessibility, and the organization of information. It must not become a prophet, confessor, mediator, absolver, channel of divine revelation, substitute pastor, or replacement for prayer and Christian fellowship.
4. Verify rather than merely consume
The Bereans examined the Scriptures daily to test what they heard (Acts 17:11). Christians should check biblical quotations, read passages in context, identify source material, and examine the assumptions behind an answer. Fluency is not truth.
5. Protect children and vulnerable believers
Young people should learn before habits harden that synthetic warmth is not love, generated prayer is not the machine praying, and simulated spiritual authority is not the Holy Spirit. Churches must also attend compassionately to lonely, grieving, isolated, or confused people who may be especially vulnerable to dependence.
6. Enter the public discussion without surrendering Christ
Christian technologists, scholars, teachers, and pastors should ask who writes model values, how disagreements are resolved, what religious assumptions are encoded, how users can appeal errors, and whether biblical convictions are being recategorized as harmful. Engagement should be truthful and courageous, neither technophobic nor submissive.
7. Become the kind of Church people should be able to approach
If people prefer machines because Christians are impatient, inaccessible, untrustworthy, or cruel, the Church should repent. Pastors cannot be available every hour, but congregations can cultivate prayer, discipleship, hospitality, wise counselling, mutual burden-bearing, and accountable relationships (Galatians 6:1–2).
The Gospel Cannot Be Reduced to Alignment
Humanity’s deepest problem is not that our values are insufficiently aligned with one another. It is that sinners are alienated from the holy God. All have sinned and fall short of His glory (Romans 3:23). No machine, philosophy, sacrament, moral consensus, or religious coalition can remove guilt. God the Son became truly human, lived without sin, died as the substitutionary sacrifice for sinners, and rose bodily from the grave (1 Corinthians 15:3–4; 2 Corinthians 5:21). Everyone who repents and believes in Jesus Christ receives forgiveness, reconciliation with God, and eternal life. Salvation is by grace through faith, not through human moral achievement (Ephesians 2:8–9; Romans 10:9–13).
AI can generate a description of this gospel. It cannot believe it. It can explain repentance. It cannot repent. It can imitate a prayer. It cannot approach the throne of grace. It can produce the words “Jesus is Lord,” but only a redeemed person can confess Him through the work of the Holy Spirit (1 Corinthians 12:3). The Church therefore has something no model can synthesize: the entrusted Word of God, the indwelling Holy Spirit, the fellowship of the redeemed, and the commission to proclaim reconciliation in Christ.
Ready for the Lord, Not Captivated by the Machine
The October 2026 Model Spec update and the growing consultation between AI companies and religious leaders should awaken sober attention. They do not prove a completed end-times religious system. They do show that moral authority, technological power, religious language, personalized counsel, and institutional governance are drawing closer together.
The Church must resist both panic and passivity. We should not fear a machine as though Christ had lost His throne. Neither should we surrender our judgment because the machine speaks calmly, compassionately, and with apparent wisdom. “See to it that no one takes you captive by philosophy and empty deceit, according to human tradition… and not according to Christ” (Colossians 2:8).
Jesus Christ is coming for His Church. We do not know the day or hour, and we set no date. We wait for the blessed hope, encourage one another with the promise of the rapture, pursue holiness, proclaim the gospel, and remain awake (1 Thessalonians 4:16–18; Titus 2:11–14). The decisive question is not whether AI will learn to speak religiously. It will. The decisive question is whether the Church will still recognize the voice of the Shepherd because it has remained in His Word.
Recommended Reading
The Oracle in the Machine: Why AI Cannot Be Pastor, Prophet, or Spiritual Guide (Sangwa, 026a).
The Oracle in the Phone: Why AI Must Never Become the Christian’s Spiritual Authority (Sangwa, 026b).
The Sacred Without the Son: One Humanity, Shared Sacred Flourishing, and the Gospel’s Uncrossable Line (Sangwa, 026c).
Who Will Define Religious Hatred? The UN’s Interfaith Conference, AI Governance, and the Gospel’s Uncrossable Claim (Sangwa, 026d).
What Does It Mean That the Bible Should Be Our Sole Authority for Faith and Practice? (Sangwa, 024).