The Use, Not the Tool

Type is using AI a sin into a search box and the suggestions that come back finish it four ways: in the Bible, Catholic, for homework, in Islam. Those are four different questions. The first asks about a book that never mentions computers. The second asks about a church whose two main documents on AI run to 362 numbered paragraphs between them. The third asks about a student’s honesty. The fourth asks a tradition whose scholars issue rulings, fatwas, on exactly this kind of question.

This essay answers the practical question a believer is asking: is it wrong for me to use this, and if some uses are wrong, which ones and why? It does not collect the documents again. The church texts on AI, quoted with paragraph numbers, are in A Pale Reflection, and the five Bible passages people cite about AI, quoted in the King James Version and read in their chapters, are in Mouths That Speak Not. Here the question is narrower: on what grounds do teachers of three traditions say a particular use could be wrong?

How the sources were handled. Every text below was opened with a public GET on 10 October 2026 and quoted from the words as served: a fatwa and an article from Egypt’s Dar al-Ifta, two answers from Darul Ifta Birmingham, a fatwa from Islamweb, the Southern Baptist Convention’s 2023 resolution on AI, an episode of John Piper’s Ask Pastor John, one paragraph of the Vatican’s Antiqua et Nova, Rabbi Daniel Nevins’s 2019 responsum for the Conservative movement’s law committee, and Rabbi Gil Student’s column on AI and homework. Desiring God’s own page refused an automated request (HTTP 403), so Piper is quoted from the Internet Archive’s copy of it. Each opinion is attributed to the person or body that published it, never to “Islam”, “Judaism” or “Christianity” as a whole, and no ruling is taken from a forum. No scripture is quoted here; where a teacher’s argument rests on a verse, the essay says so and points to the source. Every quotation is recorded in this site’s source ledger.

Christian teachers

Catholic: deception in education is named

The Catholic Church’s main text on AI is Antiqua et Nova, a doctrinal note of January 2025 approved by Pope Francis, and it is set out in A Pale Reflection with Pope Leo XIV’s 2026 encyclical. Its general line is that AI is a tool, that people stay answerable for how it is used, and that each use has to be judged in its context. On the homework question it is unusually direct. Paragraph 62 first says that passing AI off as a person should always be avoided, and then goes on:

Similarly, using AI to deceive in other contexts—such as in education or in human relationships, including the sphere of sexuality—is also to be considered immoral and requires careful oversight to prevent harm, maintain transparency, and ensure the dignity of all people.Dicastery for the Doctrine of the Faith and Dicastery for Culture and Education, Antiqua et Nova, 28 January 2025, ¶62

Note what makes the act immoral in that sentence: deceiving. The note does not call a student’s use of AI immoral; it calls using AI to deceive immoral. Its worry about classrooms that do not involve deception is put as a risk, not a sin: paragraph 81 says heavy use of AI in education could increase students’ reliance on technology, eroding their ability to perform some skills independently. And its gravest warning, that treating AI as a substitute for God is idolatry, is quoted and discussed in A Pale Reflection and, with the psalm it echoes, in Mouths That Speak Not.

Protestant: honesty first, and the pulpit as a special case

Protestants have no single teaching office, so there is no Protestant ruling, only statements by bodies and teachers. The largest is the Southern Baptist Convention, whose messengers adopted a resolution on AI in New Orleans in June 2023. Its posture is the opposite of a ban. It says the messengers want to engage these technologies from a place of eschatological hope rather than uncritical embrace or fearful rejection, and it states the moral line in one clause, encouraging users to employ AI in honest, transparent, and Christlike ways,

never seeking to willfully deceive others or take advantage of them for unjust gain or the accumulation of powerSouthern Baptist Convention, On Artificial Intelligence and Emerging Technologies, resolution adopted June 2023

The resolution builds on the 2019 evangelical statement of the convention’s Ethics & Religious Liberty Commission, which A Pale Reflection quotes at length. It also adds a reason for care that is not about any single act: it says these tools are designed with distinct values and purposes in mind and shape us in subtle, yet meaningful ways.

John Piper, the Baptist pastor and founder of Desiring God, took the question at its sharpest point in an episode of February 2025, when a pastor asked whether he could have ChatGPT draft his sermons. Piper had asked ChatGPT for an answer to a similar question in his own style, and said the result was so good his listeners would not have known. Then:

And if I had read that to you as my own, it would have been wicked.John Piper, “Should I Use AI to Help Me Write Sermons?”, Ask Pastor John, episode 2127, Desiring God, 24 February 2025

His reason is the same as the Baptist resolution’s, stated as a principle about God and his people: That is, we do not deceive. But Piper goes further than honesty in one place. Using ChatGPT for information, even for inspiration is fine, he says, just like commentaries and books. Having it write the first draft of a sermon, which the pastor then checks and adjusts, is something he says he is appalled by, because the ability to teach is a biblical qualification for the pastor and It’s your number-one job. If a pastor does it anyway, Piper says the congregation must be told. His deeper point is about worship: machines can simulate reasoning, but no artificial intelligence will ever be able to worship, so it cannot take the believer’s place before God. What happened when churches and a synagogue actually tried a machine-written service is told in Ninety-Eight Percent from the Machine.

Muslim scholars

Islamic law has a vocabulary for exactly this question, and the most formal answer opened for this essay uses it. In December 2025 Egypt’s Dar al-Ifta, answering under the name of Dr. Nazir Mohammad Ayyad, Egypt’s Grand Mufti, was asked whether it is lawful to use AI applications to get fatwas. It begins by ruling on AI as such:

Though it is permissible in essence, the ruling of Islamic law on AI is contingent on the purpose for which it is used.Dar al-Ifta al-Misriyyah, “Using AI applications to obtain fatwas”, 2 December 2025

The fatwa rests that on a Qur’anic verse about what God has made of service to humanity and on a hadith that what God left unmentioned is pardoned, and then on a legal maxim it quotes: The means take the same ruling as the goals. AI is a means. Used for something lawful it is lawful; used for something forbidden it is forbidden. That is the most precise statement in any of these traditions of the idea in this essay’s title.

Then comes the use it forbids. A fatwa, it says, needs a scholar who knows the sources, the methods of deriving rulings and the circumstances of the person asking, and AI answers may be right, wrong, or even contradictory to established rulings. Its conclusion: Based on the above, it is impermissible in Islamic law to rely on AI applications to obtain a fatwa. The word is rely. The Grand Mufti’s own article on AI and family questions, undated on the page, draws the same line and is explicit that AI has real uses, among them locating texts, summarising scholarly positions and translating sources: The issue, therefore, is not technology itself, but how it is used, for what purpose, and under whose guidance. What it cannot do is stand in for the scholar: But the ability to generate an answer is not the same as the authority to issue a fatwa.

Image-making: where Muslim scholars split

The one place where using AI can come close to a sin in itself, for some scholars, is pictures. Many classical jurists forbade making images of living beings, on the strength of hadiths about image-makers. Whether a picture generated by a prompt counts is now answered both ways by fatwa services that publish in English.

Darul Ifta Birmingham, a British institute of Islamic jurisprudence, answered in October 2024 (fatwa ID 07541, written by Maulana Yusuf Khan, checked by Mufti Mohammed Tosir Miah). It sets out the classical view that drawing full figures of living beings is not allowed, and it records a distinction from the commentators: making images in order to imitate God’s creation is far graver than making them without that intent, and Those who create idols to be worshipped are seen as committing an even more grave crime. It then states the open question plainly: with photography and now AI, scholars have debated whether these forms of image-making fall under the same category as traditional drawings. It does not issue a blanket ban. It lists what decides a case: how the image was made, how realistic it is, what it is for, and whether it carries the old dangers, which it names: The prohibition of image-making is closely tied to the fear of idolatry and the promotion of immorality. Its conclusion is conditional. The more lifelike an image of a person, the more it may be seen as imitating creation; immodest images are not allowed; and images that stay within those limits and serve education or another benefit are permissible, because an AI image is not the drawing the hadith speaks of. The institute’s site also lists a separate answer on realistic AI images, which did not load when requested on 10 October 2026, so it is not reported here. Another Birmingham answer says AI pictures of landscapes, buildings and trees raise no issue at all, because Non-living objects like mountains, rivers, houses, and inanimate landscapes have never been included in the prohibition of image-making.

Islamweb’s fatwa centre went further in March 2026 (fatwa 528754). Asked by someone who makes AI videos of insects, birds and sea creatures, it answered:

There is no religious harm in generating digital images of living beings using artificial intelligence, and this does not fall under the prohibited form of image-making.Islamweb, “Ruling on AI-Generated Images, Using Music, and Earning from YouTube”, fatwa 528754, 30 March 2026

Its reason is that such images have no physical existence and appear only on a device. The same fatwa then forbids the background music the questioner used, which is a reminder that a scholar who permits the AI part of a project may still forbid other parts of it.

These are answers to individual questions. What Muslim institutions have published on AI as a whole, from the Fiqh Academy of the Organization of Islamic Cooperation to Al-Azhar’s Grand Imam, is read statement by statement in Permissible by Default.

Jewish teachers

Jewish law handles the homework question with a concept older than writing machines: geneivat da’at, literally “theft of the mind”, creating a false impression in someone else’s mind. Rabbi Gil Student, editor of the website Torah Musings, applied it to AI in a column for the Jewish Link in January 2023, weeks after ChatGPT appeared. He reports that Rav Moshe Feinstein, in his responsa Iggerot Moshe, forbade students to copy from each other on these grounds, and extends the ruling:

The same should apply to submitting the work of AI as your own. You are misrepresenting yourself as having skills, knowledge and achievements that you do not have.Rabbi Gil Student, “Can AI Do Your Homework?”, Jewish Link, 12 January 2023

Then he adds the qualification that makes his answer different from a ban. Cheating, he says, depends on what the teacher expects, just as using a calculator was cheating when he was at school and is expected now. Students are currently expected to write essays without AI, but When that changes (I say when and not if), usage of that technology will not constitute cheating. Until then, using it against expectations is geneivat da’at. He also considers and rejects one ground: the duty to credit and thank an author does not, he thinks, apply to a machine, because We do not thank the tools and technology we use to accomplish things.

The fullest Jewish legal treatment opened for this essay is Rabbi Daniel Nevins’s responsum Halakhic Responses to Artificial Intelligence and Autonomous Machines, approved 18–0 by the Conservative movement’s Committee on Jewish Law and Standards on 19 June 2019. It is not about chatbots, which barely existed then, but it rules on the questions underneath this essay’s. On the Sabbath, it says, a Jew should not request that a smart machine initiate or complete forbidden labor unless there is a mitigating reason such as illness, while services arranged before the Sabbath may be permitted. On religious duties, a machine may help but may not stand in. A robot could fetch a prayer shawl for someone who cannot move, but the ritual itself must be done by a person for whom it is an obligation, and the same holds for acts of kindness such as visiting the sick and comforting mourners:

A machine might assist, but a human must initiate and complete this action.Rabbi Daniel Nevins, Halakhic Responses to Artificial Intelligence and Autonomous Machines, CJLS HM 182, approved 19 June 2019

Nevins is clear that this is a line, not a dismissal. Machines do not become humans, he writes, but This distinction is not to diminish their value, which can be vast. He also notes that the question is old: in the early modern period rabbis asked whether a golem could be counted in a minyan, the quorum for prayer, and Rabbi Jacob Emden concluded it could not.

The grounds, side by side

Put together, the published texts give five grounds on which a use of AI could be wrong. None of them is the use of AI as such.

• Deception. The clearest and most widely shared. The Southern Baptists, John Piper and Antiqua et Nova each name it; Gil Student names it in Jewish law as geneivat da’at. The wrong is the false impression, so the remedy each names or implies is disclosure.
• Homework and plagiarism. A case of deception, made explicit by Antiqua et Nova (“in education”) and by Student. Student ties it to what the teacher expects; the Vatican adds a separate worry, short of a sin, about skills that never get built.
• Image-making. A distinctly Muslim question, on which the fatwas opened here differ: Islamweb permits AI images of living beings without conditions beyond their content, Darul Ifta Birmingham permits them only within limits of realism, modesty and purpose, and both permit pictures of things without souls.
• Replacing prayer, ritual or counsel. Dar al-Ifta forbids relying on AI for a fatwa; Nevins says a machine may help with a commandment but not perform it; Piper says a machine cannot worship and a pastor may not hand it his teaching.
• Idolatry. The gravest ground and the rarest in practice. Antiqua et Nova warns against treating AI as a substitute for God; the Birmingham fatwa names idol-making as the root of the image ban; the Baptist resolution ends by confessing that no technology will ever usurp the sovereignty and power of God.

Where they disagree

The traditions agree more than the search suggestions imply. Every teacher read here starts from the position that the tool is permitted and the use is judged. The real disagreements are narrower, and they are worth knowing because they decide some everyday cases.

Is disclosure enough? For the Southern Baptist resolution and for Gil Student, deception is the wrong, so honesty cures it: a student whose teacher allows AI, or a writer who says what the machine did, has done nothing wrong. Piper agrees for newsletters, if the newsletter says at the top that ChatGPT wrote it. For sermons he goes further: he is appalled at the thought of a machine writing the first draft, because it hands over the task the office exists for. He still says that a pastor who does it must tell the congregation, so disclosure makes it honest; it does not stop him calling the practice appalling.

Does expectation decide? Student’s answer moves with the classroom: once teachers expect AI, using it is no longer cheating. The Catholic note’s concern about lost skills does not move in the same way, because it is about what happens to the student, not about whom the student fools. A school can change its rules; it cannot, on the Vatican’s account, change what reliance does to a mind.

Do AI pictures count as image-making? Islamweb says no, flatly, because a screen image has no physical existence. Darul Ifta Birmingham reports that scholars are still debating it, and gives a conditional yes to permission: an AI image is not the drawing the hadith speaks of, but the closer it comes to a lifelike person the more it may count as imitating creation, and it needs a purpose of benefit. A Muslim asking whether to generate a realistic portrait for fun would get a plain yes from one service and, at best, a hesitant answer from the other. Behind both stands the older view, which Birmingham sets out, that making pictures of living beings by hand is a sin even without bad intent.

Where may a machine stand in religious life? The three traditions draw the line in the same place for different reasons. Dar al-Ifta keeps the fatwa for a scholar because the scholar carries religious accountability; Nevins keeps the commandment for a person because it is that person’s obligation; Piper keeps the sermon for the pastor because teaching is his calling. None of them forbids a believer to use AI to find a text or check a fact.

The strongest objection: the use is never neutral

Here is the case against this essay’s answer, as strongly as I can make it.

“Judge the use, not the tool” sounds wise and lets everyone off. It assumes there is a person, already formed, choosing uses of a neutral instrument. But the tool forms the person. The Southern Baptist resolution itself says these tools shape us in subtle, yet meaningful ways, the Catholic note warns that heavy use erodes skills students would otherwise build, and both the evangelical statement and Pope Leo XIV’s 2026 encyclical, as A Pale Reflection records, refuse to call AI morally neutral. If the habit of asking a machine first slowly replaces asking God, a parent, a teacher or one’s own conscience, then no single use was a sin and the sum of them is exactly what every tradition here fears: a believer who has handed over thinking, counsel and finally trust. On that view, “is using AI a sin?” deserves a less comfortable answer: it is a near occasion of sin, as tempting and as habit-forming as anything the moral tradition warns about, and a believer should treat it with that suspicion rather than with a checklist of five forbidden uses.

That objection is right about a good deal. It is right that the texts read here worry about habits as well as acts, and that the Catholic and Baptist documents say so outright. It is right that the Grand Mufti’s concern is precisely the teenager who asks a chatbot before a parent or a scholar. And it is right that a list of forbidden uses can become a way of not asking what a tool is doing to you.

Where it fails is as an answer to the question asked. None of the teachers read here calls the use of AI a sin, and the one who comes closest to the objection’s spirit, Dar al-Ifta, states the opposite in so many words: permissible in essence, judged by purpose. The objection turns a warning about character into a ruling about acts, which is a move the sources do not make. What it rightly adds is that “judge the use” includes judging the pattern of use, not only each act. A believer who takes the traditions seriously should ask not only “am I deceiving anyone?” but also “what am I no longer doing because the machine does it?”

What this leaves

• No tradition read here calls using AI a sin in itself. Dar al-Ifta says so explicitly; the Baptist, Catholic and Jewish texts judge uses, not the tool.
• Deception is the common ground. Passing off machine work as your own, in homework, a sermon or a newsletter, is named as wrong by teachers in all three traditions.
• Religious authority is kept for people. A fatwa, a commandment and a sermon are each held back for a human being who carries the responsibility for it.
• Pictures are contested. Muslim scholars disagree about AI images of living beings; all of those read here permit images of things without souls.
• Habits are a real concern, not a ruling. The strongest objection shows that a use can be permitted and still worth watching.

For what the churches have published about AI, document by document, read A Pale Reflection; for the Bible passages people cite, Mouths That Speak Not; for what happened when congregations heard a machine’s sermon, Ninety-Eight Percent from the Machine. Books by believers who have written on AI at length, Christian and Jewish among them, are on the reading list.