AI Fake Scholars: Why Blind Trust Is a Trap
AI Fake Scholars: Why Blind Trust Is a Trap
Can AI Fake an Islamic Scholar?
Generative AI is forging scholars' faces and voices — and exposing how fragile our trust in religious authority has become.
Summary: Generative AI is creating an unprecedented crisis — forging the faces and voices of Islamic scholars so that words they never spoke spread virally under the banner of “authority.” In 2026, a TikTok account called “Niahajar” accumulated hundreds of thousands of followers. Its “Islamic female scholar” persona, combined with glaring errors in her Quranic recitation, raised a question no one could answer: does she actually exist? This is not an isolated case. When AI can turn any face and any voice into a spokesperson for “religious scholarship,” what is damaged is not merely an individual’s reputation but the Muslim community’s basic trust in where knowledge comes from. This article argues from the foundational principles of Islamic epistemology, drawing on the Quran and Hadith, that blind following of human beings is itself a trap to be wary of, and that only Allah’s Book and the Sunnah of Prophet Muhammad (peace be upon him) constitute unquestionable guidance.
1. A Scholar Who Does Not Exist, and a Very Real Problem
What happened on Indonesian social media is emblematic. The account “@niahajars” posted content that looked ordinary: clips of religious lectures, faith-based motivation, encouragement to pray and send blessings on the Prophet. But users gradually noticed that the “female scholar’s” recitation contained clear errors in Arabic pronunciation — and more fundamentally, no one could establish her real identity.
This is not a technical glitch. It is a textbook case of generative AI being used to fabricate religious authority. Academic research has already conceptualized the phenomenon: AI-generated content “simulates the voices of clergy, fabricates sacred symbols, and constructs emotionally resonant narratives about the ummah,” with the core harm being that it “separates technical precision from moral accountability, producing sanad-less knowledge” — knowledge with no chain of transmission.
In the Islamic tradition of knowledge, the chain of transmission (sanad) is not formalism. Whether a hadith is reliable depends first on whether the people transmitting it are reliable, whether they heard it directly, and whether the chain is continuous. This is a verification mechanism that Muslim scholars built over centuries. The AI-forged “scholar” destroys precisely this mechanism: you see a face, you hear a voice, but you cannot trace any actual human being who is accountable for the words.
2. “Do Not Follow What You Have No Knowledge Of” — A Foundational Command That Is Being Ignored
There is one verse of the Quran that is almost impossible to bypass when discussing AI-forged religious authority:
“And do not follow that of which you have no knowledge. Indeed, the hearing, the sight and the heart — about all those [one] will be questioned.” — Surah Al-Isra 17:36
Ibn Bāz, in explaining this verse, stated clearly that Allah forbids a person from speaking, judging, acting, or believing without knowledge. A person will be held accountable for his ears, his eyes, and his heart. Therefore: “Do not say ‘I heard’ unless based on clear evidence; do not say ‘I saw’ unless based on clear evidence; do not believe something in your heart unless based on clear evidence.”
The application of this verse today has become extraordinarily concrete. When you scroll past a video of a “scholar” delivering a moving message, your ears hear it and your eyes see it. But do you “know”? Do you know this person actually exists? Do you know these words truly came from him? Do you know whether he is qualified to speak on this matter? If the answer is no, then following it is precisely what Allah has prohibited.
This is not a call for Muslims to stop listening to scholars. It is a call to recognize that the precondition for following (taqlīd) is confirming the authenticity and reliability of the one being followed. AI forgery severs this precondition. What you are following is a generated face and a synthesized audio track — not a human being who can bear moral and legal responsibility.
3. The “Deification” of Human Beings Is Dangerous in Itself — Even Scholars
A core point deserves serious treatment: no human being should be turned into an idol who “cannot err.” However learned or pious a scholar may be, he remains human. Humans make mistakes, misjudge, and may even deviate from the right path. If people, because of someone’s past reputation, accept everything he says without discernment, then that attitude itself contradicts the Islamic understanding of the Oneness of God (Tawḥīd).
Ibn Bāz, in another admonition, cited the Prophet’s stern warning:
“Do not praise me excessively as the Christians praised the son of Mary. I am only a servant, so say: the servant of Allah and His Messenger.” — Sahih al-Bukhari
Ibn Bāz explained that this hadith prohibits “praise that exceeds the facts” — such as claiming the Prophet knows the unseen, or that he can be invoked alongside Allah. He further noted that even excessive praise of a living scholar is dangerous, because it “may cause the one praised to become enamored with himself, thereby cultivating arrogance and self-conceit.”
There is a subtle but crucial distinction here: respecting scholars is an Islamic tradition, but treating a scholar as an “unquestionable authority” is a betrayal of the Islamic understanding of knowledge. The Muslim community is required to verify, to ask, to confirm. When AI can easily forge someone who “looks like a scholar” saying anything at all, uncritical trust is no longer “respect” — it is credulity, and credulity, in the vocabulary of the Quran, falls under the category of “following what you have no knowledge of.”
4. “Refer It to Allah and the Messenger” — The Only Exit in Dispute
Surah An-Nisa, verse 59 establishes a clear hierarchy of authority:
“O you who have believed, obey Allah and obey the Messenger and those in authority among you. And if you disagree over anything, refer it to Allah and the Messenger, if you should believe in Allah and the Last Day. That is the best [way] and best in result.” — Surah An-Nisa 4:59
The official commentary of Turkey’s Diyanet on this verse points out a key detail: when the verse mentions “those in authority” (ulu’l-amr), it does not repeat the word “obey.” The commentary notes that this grammatically implies that the commands of those in authority are only to be obeyed when they are in conformity with the teachings of Allah and the Messenger. The hadith expresses it even more directly: “There is no obedience to a created being in disobedience to the Creator.”
This means that even a legitimate, real, and recognized religious or political authority is obeyed conditionally. What, then, of a forged authority? What of an “AI scholar” whose very existence you cannot confirm?
The danger of AI-forged religious content lies precisely in the fact that it bypasses this hierarchy. It gives you no opportunity to test “does this person actually exist” or “do his words conform to the Quran and Sunnah.” It exploits emotional reactions: a moving speech, a pious face, a sentence that “sounds very Islamic.” But Islamic epistemology demands verification, not emotional impulse.
5. Islamic Institutions Have Already Sounded the Alarm
This is not a purely academic problem. Egypt’s Dar al-Ifta, when announcing its Tenth Global Fatwa Conference (August 2025), explicitly set the theme as “Cultivating the Wise Mufti in the Age of Artificial Intelligence.” Grand Mufti Nazir Mohamed ‘Ayyad warned at the press conference that some are attempting “to use AI tools to spread erroneous fatwas that deviate from the uprightness and tolerance of Islamic law.”
At the same time, academic research on AI-generated religious content notes that the traditional Islamic verification tools — the chain of transmission (isnād) and verification (tabayyun) — remain applicable in principle, but in AI-mediated environments “their application becomes increasingly complex because authority is no longer easily traceable.”
A 2026 study published in Al-Irfan examined contemporary Muslim scholars’ positions on AI and reached a representative conclusion: mainstream scholarly positions do not reject AI technology itself, but judge it by its purposes and consequences. Scholars strongly oppose applications involving deception, deepfake manipulation, and “unauthorized religious rulings.” The study noted that contemporary Islamic discourse is increasingly inclined toward a “human-AI collaboration model,” in which AI is an auxiliary tool and ultimate moral and legal judgment must remain exclusively human.
In other words, the question is not “can AI be used to spread Islam,” but who is accountable for the output. When a forged “AI scholar” utters wrong or misleading words, there is no real person to bear that responsibility. The Islamic scholarly tradition is built on “accountable transmission of knowledge” — transmitters have names, lives, deaths, and make both right and wrong judgments. AI-generated “scholars” have none of these.
6. The Only “Guiding Light” Is Not Any Human Being
The core argument needs to be presented in full: Prophet Muhammad (peace be upon him) is the only guiding light and the last Prophet; the Quran and Hadith are the only sources of guidance. Beyond that, no one — whether called a scholar, a philosopher, or anything else — should be set up as an unquestionable standard.
Within the Islamic theological framework, this is not an extreme statement — it is a necessary corollary of Tawḥīd. Allah commands the Prophet to announce in the Quran:
“Say: I am only a man like you, to whom has been revealed that your god is one God.” — Surah Al-Kahf 18:110
The Prophet’s “human” nature is explicitly emphasized here. His uniqueness lies not in possessing some superhuman knowledge or power in himself, but in the fact that he received revelation from Allah. Guidance comes from Allah, is conveyed through the Prophet, and is recorded in the Book. The source of the chain is Allah — not any human being.
This explains why “unlimited trust” in any person is untenable in Islam. The value of a scholar lies in his effort to learn and convey knowledge, but his conveyance must always be weighed against the scale of the Quran and Sunnah. When AI forges a scholar saying something, that statement does not gain authority because it “appears to come from a scholar.” Its authority depends solely on whether it conforms to the Quran and Sunnah.
7. So What Should Muslims Do?
This is not a call for Muslims to reject all digital content. It is a call to restore “verification” as an act of faith.
First, ask about the source. Can you find the person in the video in real life? Are his identity, institution, and academic background verifiable? If not, do not treat his words as religious evidence and do not spread them. AI-forged “scholars” work precisely because most people skip this step.
Second, test the content. Even if the speaker really exists, his words must still be understood in light of the Quran and Sunnah. No scholar’s statement is “exempt from inspection.” Ibn Bāz, in explaining “do not follow what you have no knowledge of,” emphasized exactly this: knowledge brings responsibility. If you spread a statement, you are responsible for it.
Third, reject emotional manipulation. The greatest weapon of AI-generated religious content is emotional resonance. A tear-jerking speech, a “pious” expression, a slogan that sounds unassailable — none of these constitute religious evidence. Research has already noted that Islamic epistemological tools may be insufficient when confronted with the “emotional dimension” of AI content. But precisely because of this, Muslims need to consciously separate emotional reactions from knowledge-based judgment.
Fourth, place ultimate trust where it belongs. The Quran has not been forged. The chains of hadith transmission have been tested for over a thousand years. The Prophet’s teachings are clear. When every person and every screen is telling you “listen to me,” the Muslim is required to return to the unchanging source.
8. The Numbers: The Internet Is Now a Primary Source of Religious Information
The data is unambiguous, even if incomplete. Pew Research Center estimates that approximately 40% of young people in Muslim-majority countries obtain religious information from the internet. An Arab Barometer survey found that 60% of 18- to 29-year-old Muslims in the Middle East are both “heavy internet users” and “highly engaged” on social media. Among British Muslims, roughly 30% use social media “a great deal” for religious information.
A 2026 study of Muslim students found that 73.3% engage with Quran-Hadith content via digital media at least three times per week. YouTube was the dominant platform (35.0%), followed by TikTok (25.8%), Instagram (17.5%), and dedicated Quran/tafsir/hadith apps (15.0%). Audiovisual platforms were used by 78.3% of respondents, indicating a strong preference for visual, concise, algorithmically recommended religious content.
These figures describe access, not depth. And that distinction is everything.
9. The Form of the Medium Is the Problem
TikTok’s architecture does not merely deliver religious content. It reshapes what religious content can be. A major 2026 analysis of Indonesian Islamic TikTok content found that 58.3% of videos were under 60 seconds, with only 123 out of 1,351 exceeding three minutes. The study concluded that “religious authority is increasingly measured in seconds of retained attention rather than depth of doctrinal exposition — a shift with significant implications for how piety is taught, performed, and evaluated in digital public space.”
This is not a neutral change in format. It is a transformation in what counts as religious knowledge. A 60-second video cannot establish the context of a Quranic verse, cannot trace the chain of a hadith, cannot distinguish between a general principle and a specific ruling, and cannot demonstrate the qualifications of the speaker. What it can do is provoke an emotional response. And emotional response, as the preceding section argued, is precisely what AI-forged content is optimized to generate.
The same study documented exponential growth in Islamic short-form content: from 2 videos in 2020 to 610 in the first quarter of 2026 alone. The volume is overwhelming. The depth is not.
10. The Illusion of Learning Without the Substance
Here is the most dangerous misconception: that consuming Islamic content on social media constitutes “studying Islam.”
The 2026 study on Muslim students captured this paradox precisely. While 73.3% engaged with Quran-Hadith content digitally multiple times per week, 74.2% still regarded conventional learning as the most authoritative source for deeper understanding. Students know, at some level, that the feed is not the classroom. Yet the feed is where they spend their time.
This is the definition of a weak foundation. A person who watches a hundred short videos about Islam but has never opened a tafsir, never read a hadith collection with commentary, never sat with a teacher to work through a text, has exposure without literacy. They recognize names and slogans. They cannot evaluate a claim. They cannot distinguish a sound argument from a rhetorically effective one. They cannot ask: Where did this come from? Who said it? Is the chain sound? Does it contradict something established?
These are not advanced skills. They are the minimum requirements of Islamic epistemic responsibility. And they are precisely the skills that short-form content does not teach, because the medium has no room for them.
11. The Vulnerable Audience Meets the Fabricated Authority
The preceding sections described AI-generated “scholars” whose faces and voices are forged. It is essential to be explicit about why this works.
An audience with strong foundational literacy would ask: Who is this person? What is her chain of transmission? Where did she study? Can I verify this? These questions are not skepticism. They are the Islamic method. The entire science of hadith criticism (jarḥ wa taʿdīl) exists because the early community understood that claims about religion must be verified, and that verification requires traceable human accountability.
An audience with weak foundational literacy asks a different question: Does this feel right? Does this speaker seem pious? Does this message resonate with what I already believe? These are not verification criteria. They are receptivity criteria. And AI-generated content is designed precisely to satisfy them.
The academic literature on AI and Islamic epistemology has already identified the mechanism. One 2026 study proposed an “Epistemic-Boundary Model,” arguing that AI must be restricted to assistive functions — retrieval, clustering, reference tracking — while verification of transmission chains and legal deduction remain exclusively human domains. The study warned that AI “risks reducing the complexity of Hadith epistemology to a purely computational process while simultaneously obscuring the foundations of Islamic scholarly legitimacy.”
But this warning presupposes that the human audience knows what those foundations are. A community that has stopped studying the sources cannot recognize when the foundations are missing. It sees a face, hears a voice, and receives a message. It does not see the absence of the chain, because it never learned that the chain matters.
12. There Is No Mechanism to Label the Fake
The final and most concrete failure is institutional. Even if an individual Muslim wanted to verify the authenticity of a viral “scholar” video, there is no large-scale, accessible, systematic mechanism to do so.
The volume of content makes manual review impossible. As one IEEE paper on Islamic misinformation detection noted, initial attempts to establish organizations that manually review published news “mostly failed due to the overwhelming volume of information that needs to be verified, as this process requires extensive time, effort, and extremely high costs, yet it still cannot cover all the claims being published.” The paper proposed a technical solution — an evidence-aware detection framework using CAMeLBERT and external evidence retrieval — but this remains a research prototype, not a deployed consumer tool.
On the platforms themselves, there is no dedicated label for “AI-forged religious authority.” A deepfake of a scholar speaking words he never said is not necessarily categorized as “misinformation” by automated systems, because the systems are not trained to recognize religious authority markers. And even if it were flagged, the platform’s response would be generic — a warning label, perhaps — not an Islamic scholarly assessment of what was fabricated and why it matters.
The Islamic tradition has a mechanism: the isnād, the chain of transmission, the requirement that knowledge be traceable to a reliable human source. No digital platform has replicated this mechanism at scale. The feed has no sanad.
13. The Foundation Must Be Rebuilt
This is not a call to abandon digital tools. It is a call to recognize that no tool can substitute for the foundation.
The Quran and Hadith are not “content.” They are the source and criterion of all content. A Muslim who has not studied them — who has not sat with them, wrestled with them, learned from those who learned from those who learned — has no criterion by which to judge what appears on a screen. They are, in the precise sense, defenseless.
Ibn Bāz’s explanation of Surah Al-Isra 17:36 (“Do not follow what you have no knowledge of”) applies here with devastating force. The verse holds the ears, eyes, and heart accountable. A Muslim who watches a forged scholar and shares the video has used their ears and eyes to transmit something they did not know. The accountability is not merely for the original forger. It is for the transmitter who failed to verify.
The Prophet’s warning is equally direct: “It is enough falsehood for a man to speak of everything that he hears” (Sahih Muslim). This hadith was narrated fourteen centuries before the algorithmic feed. Its application has never been more urgent.
14. Conclusion
The AI-forged scholar is a symptom. The disease is a community that has, in large numbers, outsourced its foundational literacy to platforms that were never designed to cultivate it. The internet provides access to more Islamic material than any generation in history. It also provides access to more fabricated authority than any generation in history. Without the foundation to distinguish between them, access becomes a liability.
The Quran and Hadith remain. They have not been forged. They have not been algorithmically optimized. They require effort — the effort of study, the effort of verification, the effort of sitting with teachers and texts. That effort is not optional for a community that claims to follow the final Prophet. It is the minimum condition of being able to tell the difference between guidance and a generated imitation of it.
Allah’s pairing, in the Quran, of “following what you have no knowledge of” with the accountability of the ears, eyes, and heart is not accidental. It reminds every Muslim that the most precious tool you possess is discernment. And the first lesson of discernment is to acknowledge that no human being, however authoritatively packaged, can stand in for you before Allah’s words.
The Prophet is the guiding light; the Quran and Sunnah are the foundational reference. Beyond that, everything must be tested — including the most pious-looking faces and voices.