Report
A September 30 conference presentation separates automated production from religious authority. Its one-account case study also makes the limits of the evidence essential to the story.
Researchers at Indonesia's UIN Walisongo presented a study of AI-assisted Islamic outreach on September 30, raising a question that is easy to lose in discussion of automated religious content: who remains answerable for the message? The university's report places the presentation at the International Da'wah Conference in Yogyakarta. Da'wah here concerns communicating and teaching Islam. Source: walisongo.ac.id
Fitri Ariana Putri and Usfiyatul Marfu'ah examined one social-media account through an exploratory qualitative approach, online observation and critical discourse analysis. According to the university, their conclusion distinguishes production from authority: AI carried out the creator's instructions, while religious authority remained with the creator. They identified verification of religious knowledge as an important unresolved responsibility.
The university expressly limits the findings to the case studied. AI Faith Monitor inspected its complete report, but has not obtained the underlying paper, observation records or a response from the account's operator. This is reporting about a conference presentation, not an independent finding of misconduct or evidence about all Muslim creators. The distinction matters because an institutional summary can establish what researchers say without establishing every conclusion they reached.
There are at least three different questions inside the word verification. Does an attributed passage actually exist? Does the explanation preserve its context? And is the person applying it entitled, within the relevant tradition, to offer that kind of guidance? This division is our analytical framework, not a result measured in the Indonesian case.
A hypothetical editor could answer the first question successfully and still fail the second. A quotation may be authentic while a shortened video omits the qualifications that make it intelligible. The third question concerns the responsibilities surrounding advice, rather than the technical accuracy of the quotation alone.
This also explains why a useful production tool need not become a religious authority. Transcribing an already reviewed lecture, proposing headings and answering an individual's question are different activities. An institution considering them should specify which activity it is permitting before treating one successful demonstration as approval for the others.
The new presentation sits alongside an earlier, affirmative argument for using AI with human oversight. In a July 29 report, the Indonesian Ulema Council, MUI, described digital-research official Ismail Fahmi encouraging Islamic organizations to use AI in outreach and religious literacy. He made the case during a workshop organized with BAZNAS. Source: mui.or.id
His practical example was shortening a recorded sermon into material suitable for publication. In MUI's account, the human publisher must still check the content and meaning. Fahmi also advocated AI literacy and critical thinking, rather than handing over the process of understanding. These are his recommendations, not independently demonstrated outcomes of a training program.
That position is useful context because caution need not mean opposition to every application. If a tool saves editing time and the saved time is used for careful review, assistance and accountability can coexist. Whether they actually do coexist must be checked in the particular workflow; neither the label human-reviewed nor the presence of a respected institution proves it.
Egypt's Dar al-Ifta sets a firmer boundary around a specific task. A ruling published December 2, 2025, under Dr. Nazir Mohammad Ayyad's name allows AI use in principle according to its purpose, but rejects reliance on AI applications for obtaining a fatwa. The explanation stresses the qualifications of a mufti and the need to consider the questioner's changing circumstances. Source: dar-alifta.org
This is an identifiable institution's religious judgment. It should not be expanded into a declaration that every Muslim authority rejects every AI tool. Nor should the ruling be treated as a laboratory comparison of all systems available today. Its central concern is the kind of responsibility involved in giving Islamic legal guidance.
The comparison clarifies the news without collapsing different institutions together. Encouraging a supervised editing aid and refusing to delegate a religious ruling address different tasks. A generic question about whether religion accepts AI obscures that difference.
A separate technical account helps explain why checking cannot be reduced to how certain an answer sounds. In September 2025, OpenAI published an explanation of research arguing that ordinary training and evaluation incentives can reward guessing over acknowledging uncertainty. It distinguishes correct answers, errors and abstentions, rather than treating every unanswered question as equivalent to a false answer. Source: openai.com
That is the developer's explanation of its research, not an independent evaluation of religious teaching or the account studied in Indonesia. It nevertheless supplies a relevant technical caution: a smoothly completed answer is not its own evidence. A system that declines an unsupported claim can be more useful than one that supplies a persuasive fabrication.
For a religious publisher, the practical application is to allow material to remain unpublished when its basis cannot be established. This recommendation follows from the information problem; it is not a claim that any named creator has ignored such a safeguard.
James 3:1–12 begins with the stricter accountability of teachers and then develops images of small things with large consequences: a bit, a rudder and a spark. Its warning concerns human speech, including the contradiction of praising God while cursing people made in God's likeness. It is not a prediction about language models. James 3:1-12 (NIV)
Applied to Christian publishing, the passage gives a reason to retain responsibility when production is delegated. The church that distributes a study guide remains responsible for what it teaches, whether the first article came from a staff member, a volunteer or software. An AI label can disclose a production method; it cannot carry the obligation to correct an error.
This Christian application does not decide questions of Islamic authority. It does invite Christians to learn respectfully from the specific distinction the new presentation raises. Before releasing religious material, identify the person who can explain its sources, answer challenges and issue a correction. Faster production is beneficial only when the community can still find someone responsible for the words.
UIN Walisongo: IDACON presentation on AI and religious authority
Reports the complete university account of a September 30 presentation; underlying paper, observation records and creator response not obtained. One-account exploratory study is not generalizable. Older Muslim guidance and technical research are explicitly historical context. Christian application is editorial reasoning.
Analysis / Faith & religious life
Claude’s new anti-abuse policy raises questions about machine consciousness, corporate authority and the role religious communities should play in shaping AI.
Published 2026-10-10 · Policy announced October 8, 2026; effective November 12, 2026
News brief / Faith & religious life
The adoption is confirmed by Kazakhstan’s presidential office and reporting from Astana. The full text still needs scrutiny before its practical protections can be assessed.
Published 2026-10-07 · October 7, 2026 meeting and correspondent reports; presidential account checked October 7
News brief / Faith & religious life
A four-country conference sets out principles for AI and regional cooperation. Its commitments leave implementation—and proof that safeguards work—as the next questions.
Published 2026-10-07 · Resolution October 7, 2026 · Indonesian account October 7; Brunei account October 8 local time