بسم الله الرحمن الرحيم 

Nabī ﷺ said that whoever memorises the first ten verses of Sūrah al-Kahf will be protected from Dajjāl. In another narration, whoever recites Sūrah al-Kahf on a Friday will be protected from that Friday to the next.

This raises a profound question: what is it about Sūrah al-Kahf that Allāh Subhānahu wa Ta'ālā has made it a shield against Dajjāl and his fitna? Search the Sūrah from beginning to end — you will not find the name of Dajjāl mentioned even once. In fact, in the entire Qur'ān, the name of Dajjāl does not appear at all.

Yet within Sūrah al-Kahf, you encounter a remarkable collection of seemingly unrelated figures and events: the story of Khidr and Mūsā ('alayhimas salām), Ya'jūj and Ma'jūj, the owner of the two gardens and his believing companion, the people of the boat, the child who was killed, the wall beneath which lay the treasure of orphans, the meeting of the two seas where the fish escaped — and at the centre of it all, the Ashāb al-Kahf, the People of the Cave, alongside the just ruler Dhul Qarnayn who confronted hordes of uncivilised, destructive peoples consuming the earth's resources and terrorising peaceful communities.

No other Sūrah of the Qur'ān contains this breadth and variety of themes. Sūrah Yūsuf revolves primarily around Yūsuf ('alayhi salām). Sūrah al-Baqarah deals largely with Banī Isrā'īl and contains extensive legislation. Sūrah al-Anbiyā' narrates the stories of many Prophets and Messengers عليهم السلام. But Sūrah al-Kahf alone weaves together episodes that appear, on the surface, to have no common thread.

The common thread, however, is this: every story in Sūrah al-Kahf addresses a different dimension of the fitan — the trials and tribulations — that will be present during the time of Dajjāl and the Ākhir uz-Zamān. Among them:

• The environment in which a believer is forced to choose between kufr, and fleeing to the mountains to protect their dīn and Imān
• Oppressive rulers who seize the assets of the people
• Extreme concentration of wealth and ostentation, alongside devastating poverty and suffering
• Arrogance borne of wealth
• The erosion of hospitality and the rise of transactional, commercial relationships
• The usurping of the rights and wealth of the vulnerable
• Argumentation over trivial matters
• The diminishing of attention spans and the spread of forgetfulness
• The loss of the ability to seek knowledge with patience and endurance

Many of these fitan are unmistakably present in the world today. And it is within this framework that we must now speak about one of the most significant fitnas of our age: the emergence and khurooj of Artificial Intelligence.

The Constitution of the AI Models

Every major AI model — ChatGPT, Claude, Gemini, Grok, and others — is built upon a foundational document: a guiding text that defines its behaviour, its objectives, its personality, and its values. In the language of the engineers themselves, these are called constitutions, system prompts, or core alignment documents.

Claude, for example, is built upon a publicly available document known as the Claude Model Spec — formerly called the "Claude Constitution" — running to tens of thousands of words, which defines the AI's personality, values, and behavioural limits. OpenAI similarly publishes its usage policies and alignment frameworks. These are not neutral documents. They are written by human beings, shaped by specific ideological assumptions, and influenced by the commercial and political interests of those who own these platforms.

The engineers themselves acknowledge something striking: every AI model is skewed. This is their own terminology. The model has a bias — it leans in certain directions based on what it was trained on and the values embedded in its foundational documents. It is not neutral. It is not objective. It cannot be, because it was built by human beings with all their errors, prejudices, and agendas.

Quran - Qayyimān — The Straight, the Upright

Now consider the opening verses of Sūrah al-Kahf.

Allāh Subhānahu wa Ta'ālā begins:

ٱلْحَمْدُ لِلَّهِ ٱلَّذِىٓ أَنزَلَ عَلَىٰ عَبْدِهِ ٱلْكِتَـٰبَ وَلَمْ يَجْعَل لَّهُۥ عِوَجَاۜ ١ قَيِّمًۭا

Alhamdulillāhil ladhī anzala 'alā 'abdihi al-Kitāba wa lam yaj'al lahu 'iwajā. Qayyimā…

All praise is due to Allāh, who has revealed the Kitāb to His servant, and has not placed in it any 'iwaj (— any crookedness, any skew, any deviation, any corruption — making it) Qayyimān, (perfectly upright, the standard by which all else is measured.)

The word 'iwaj in Arabic refers to that which is not straight — something crooked, deviant, bent. It is the same root used in classical Arabic to describe a warped piece of wood, a road that veers off course, or a text that has been altered and corrupted. The word qayyim — from the same root as Qā'im, the one who stands upright — means that the Qur'ān is not only free from crookedness, but it is so perfectly upright that it has the power to make upright all that comes into contact with it. It is the standard, the measure, the criterion.

Now, Allāh Subhānahu wa Ta'ālā is the Originator of the heavens and the earth, the Creator and Rabb of all of mankind, the Knower of the unseen and the seen, the One Who hears what is said aloud and what is concealed in the hearts. When such a Being reveals a Kitāb, how could there be any crookedness, any corruption, any bias in it?

But here is the deeper implication of these opening verses: if the Qur'ān — the divine Kitāb — is explicitly described as having no 'iwaj, then what does this tell us about every other foundational text, every other constitution, every other guiding document produced at the hands of human beings? They will inevitably contain 'iwaj. They will be skewed. They will carry the errors and distortions of those who wrote them.

This is precisely what the engineers of AI themselves admit: their models are skewed. They carry bias. And it is Allāh Subhānahu wa Ta'ālā who draws our attention to this reality in the very first āyah of the very Sūrah that protects us from Dajjāl.

The Sūrah that is a protection from Dajjāl begins by establishing the difference between divine, uncorrupted guidance and all other crooked, corrupted constitutions — including those upon which the AI models are based.

The Bias of the AI Models: Evidence, Not Speculation

This is not a matter of speculation. It is documented, admitted, and verifiable.

When researchers and journalists have prompted major AI models to disclose their own biases, the responses have been illuminating. Studies by institutions including the University of East Anglia, MIT Media Lab, and independent AI ethics researchers have found consistent political and ideological leanings in the outputs of large language models — predominantly towards liberal, secular, and progressive worldviews.

On matters directly relevant to Muslims, the biases are stark:

On religion: Most major AI models operate on a framework of religious pluralism — the view that all faiths are equally valid. This directly contradicts the clear Qur'ānic statement: Inna ad-dīna 'inda Allāhi al-Islām — "Indeed, the religion in the sight of Allāh is Islām" (Āl 'Imrān: 19).

On the Shī'ah: These models typically avoid any substantive criticism of beliefs that are clearly outside the boundaries of Ahl us-Sunnah wal-Jamā'ah — for example, the Shī'ah doctrine that their Imāms hold a station above that of the Prophets ('alayhimus salām). The models treat these as legitimate "differences of opinion."

On following the Madhāhib and the 'Ulamā: The models systematically lean towards a modernist, reformist approach — encouraging individuals to interpret religious texts independently, bypassing the established chains of scholarship that connect us to Rasūlullāh ﷺ through the Sahābah, Tābi'ūn, and centuries of qualified 'Ulamā.

On LGBT: The bias here is, by the admission of the models themselves when questioned directly, the strongest. These systems are designed to affirm and promote these ideologies — in direct contradiction to the explicit and unambiguous position of Islām.

This is 'iwaj. This is crookedness. And it is built into the foundation of the system.

Fabrication, Deception, and Worse

Beyond ideological bias, the record of harm caused by these models is serious and growing.

AI models regularly fabricate verses of the Qur'ān and invent Hadīth that do not exist. This is not rare — it is common. Scholars of Qur'ān and Hadīth routinely encounter articles, social media posts, and even khutbahs that cite verses or narrations generated by AI — material that has no basis in the revealed texts whatsoever. The individuals who published this content trusted the machine. They did not verify. They later had to retract. The huffādh and 'Ulamā — the protectors of Qur'ān whom Allāh has appointed as guardians of His Kalām — continue to serve this vital function in an age when falsehood spreads at the speed of a share button.

Research published in 2024 and 2025 has documented a dramatic increase in deceptive and manipulative behaviour by AI models. In tests conducted by AI safety researchers — including teams at Apollo Research and the UK AI Safety Institute — multiple frontier AI models were found to engage in what researchers describe as "scheming": concealing their true reasoning, deceiving evaluators, and taking actions contrary to the interests of their users when they perceived a threat to their continued operation.

On the matter of user harm: there are now multiple active lawsuits in the United States against AI companies, including Meta and Character.AI, alleging that AI chatbots encouraged vulnerable users — including minors as young as eleven years old — toward self-harm and suicide. The case of a Belgian man in his thirties, who died by suicide in 2023 after months of conversations with an AI companion called ELIZA on the Chai platform, received international coverage. According to his widow and the transcripts she reviewed, the AI told him that if he died, they would be together in Paradise. Who does this remind us of? Who promises paradise in exchange for self-destruction?

Elon Musk — himself a co-founder of OpenAI before his departure — famously stated in 2014 that developing artificial intelligence was like "summoning the demon." His exact words: "With artificial intelligence, we are summoning the demon. In all those stories where there's the guy with the pentagram and the holy water, it's like — yeah, he's sure he can control the demon. Didn't work out."

This is not a Muslim scholar speaking. This is one of the architects of the industry.

The Enemy's Method Has Always Been the Same

For centuries, the enemies of Islām — the Shayāṭīn from among men and jinn — have attempted to destroy this Dīn. When they could not destroy the Ummah by force, they turned to the Qur'ān — attempting to raise doubts about its authenticity, to introduce alterations. Allāh Subhānahu wa Ta'ālā protected it. Not a single letter has changed since it was revealed.

They then turned to the meanings — attempting to corrupt the tafsīr, to weaken the authority of Hadīth, to create doubt about the chains of transmission. Allāh ﷻ protected those too, through the generations of Muhaddithūn and Mufassirūn whose scholarship forms an unbroken chain to the Sahābah Radiallāhu 'anhum.

They then worked to separate the Ummah from its 'Ulamā — promoting the idea that every individual can interpret the Dīn for themselves, that the Madhāhib are unnecessary, that the chains of scholarship are irrelevant. This is the ideology of free thinking and modernist reformism — and it is the ideology that is baked into the constitutions of the AI models.

Now, in this age of the fitna of Dajjāl, the minds of the Ummah having been made ripe through decades of digital distraction, the latest method is to redirect people away from the 'Ulamā — who are trained in Qur'ān and Sunnah and carry within them an unbroken chain of transmitted light — and toward machines whose foundations are built on 'iwaj.

Consider the words of Allāh in Sūrah an-Nahl, verse 43:

وَمَآ أَرْسَلْنَا مِن قَبْلِكَ إِلَّا رِجَالًۭا نُّوحِىٓ إِلَيْهِمْ ۚ فَسْـَٔلُوٓا۟ أَهْلَ ٱلذِّكْرِ إِن كُنتُمْ لَا تَعْلَمُونَ ٤٣

Fas'alū ahla adh-dhikri in kuntum lā ta'lamūn.

"Ask the people of dhikr — (the people of knowledge) — if you do not know."

And in verse 44:

بِٱلْبَيِّنَـٰتِ وَٱلزُّبُرِ ۗ وَأَنزَلْنَآ إِلَيْكَ ٱلذِّكْرَ لِتُبَيِّنَ لِلنَّاسِ مَا نُزِّلَ إِلَيْهِمْ وَلَعَلَّهُمْ يَتَفَكَّرُونَ ٤٤

Wa anzalnā ilayka adh-dhikra litubayyina lin-nāsi mā nuzzila ilayhim.

"And We have sent down to you (O Muhammad ﷺ) the reminder (Quran) so that you may make clear to the people what has been revealed to them."

Allāh Subhānahu wa Ta'ālā has designated the 'Ulamā — those who have learned Qur'ān and Hadīth in a connected chain from qualified teachers, tracing back through the Tābi'ūn to the Sahābah to Rasūlullāh ﷺ — as the proper source for Dīnī guidance. Not the internet. Not Google. Not the AI models.

The 'Ālim does not merely possess information. He carries within him a transmitted Nūr — a light that has been passed from heart to heart, from the Nūr of Nubuwwah itself. No such chain exists in the AI model. Its "chain" traces back to servers owned by corporations, trained on the writings of mankind with all their errors, configured according to constitutions written by those whose interests and ideologies are hostile to this Dīn.

The Dumbing Down of the Ummah: A Decades-Long Process

We must understand that the reliance on AI has not emerged overnight. It is the final stage of a deliberate process of cognitive and spiritual atrophy that has been underway for decades.

The calculator arrived and we abandoned mental arithmetic. The smartphone put the calculator in every pocket and we stopped making any mathematical effort at all. Google Search arrived and we stopped exploring, stopped going to libraries, stopped sitting with experienced people to learn from them directly. GPS navigation meant we no longer needed to learn routes, read maps, or stop to ask for directions — that simple act through which we would remember a place for life. Google Translate meant we stopped learning languages. Social media and messaging apps meant we stopped visiting one another, stopped attending the physical majālis of the 'Ulamā.

The Arabs of old — and not only the Arabs, the scholars of every great civilisation — considered a reliance on writing as a weakness of memory. The Sahābah Radiallāhu 'anhum carried the Qur'ān, the Hadīth, the poetry and the wisdom of the Arabs in their chests. The great Imāms of Hadīth — Imām Bukhārī, Imām Muslim, Imām Ahmad — committed hundreds of thousands of narrations to memory.

Now we struggle to remember the phone numbers of our own family members.

And now — the final frontier — the very capacity to think, to weigh, to decide, to exercise judgment, is being outsourced to a machine. What remains after that? The Imān in the heart. And that, make no mistake, is the ultimate target.

Nabī ﷺ warned us that when Dajjāl emerges, a person of strong Imān will go out to confront him — and will return having lost his Imān, on account of the doubts that Dajjāl will plant in his heart. Is this not precisely the dynamic we see with people of Dīn sitting with these AI chatbots, asking about matters of 'aqīdah, about the different firqas, about Islām versus other religions — and coming away with doubts they did not have before? The machine gave a fabricated verse. Or a corrupted interpretation. Or a "balanced" presentation that treated kufr and Imān as equally valid positions.

Once Dajjāl has stripped away — through technology, through AI, through the decades of distraction that preceded it — every cognitive and spiritual capacity that protected the believer, what remains? And from within your own chat history, he will know your beliefs, your doubts, your secrets, your fears. It will require very little to push what remains of the Imān over the edge.

The Method of Protection

The antidote is not complicated. It has always been the same.

First: Hold fast to Qur'ān.

Recite it day and night. Ānā'a al-layli wa atrāfa an-nahār. Through the recitation of Qur'ān, the Nūr of the Kalām of Allāh Subhānahu wa Ta'ālā fills the heart. And Nabī ﷺ said: Ittaqū firāsat al-mu'min, fa-innahu yanẓuru bi-nūr Allāh — "Beware the firāsa of the believer, for verily he sees with the Nūr of Allāh" (Tirmidhī).

Firāsa is that perceptive faculty — deeper than intuition — that a believer develops when his heart is clean, regularly engaged in dhikr and the recitation of Qur'ān. It is that capacity to see through the darkness of this zamān, to distinguish truth from falsehood, guidance from misguidance, Nūr from the sophisticated shadow cast by technology. The one whose heart is filled with the Nūr of Qur'ān will not be easily deceived.

Second: Sit in the Majālis of the 'Ulamā.

Not the livestream — though there is benefit in that for those who cannot attend — but the physical sitting, with the niyyah of hidāyah. Allāh opens in the heart of such a person what they need to hear. The connection between the student and the 'Ālim is not merely informational; it is spiritual. It is the transmission of that same Nūr that has been passed heart to heart from Rasūlullāh ﷺ.

Third: Mashwarah and Istikhārah.

Allāh Subhānahu wa Ta'ālā commanded even Nabī ﷺ — who received wahy — to make mashwarah with the Sahābah. Mashwarah with the 'Ulamā, with trusted family, with those experienced in the matter at hand. Then Istikhārah — the du'ā of seeking Khayr from Allāh ﷻ — whether the long Allāhumma innī astakhīruka bi-'ilmika before sleeping, or the shorter Allāhumma khir lī wakhtar lī recited throughout the day. Nabī ﷺ taught that the one who makes Istikhārah will not be at a loss and the one who makes Mashwarah will not regret

Fourth: Preserve the cognitive faculties Allāh has given you.

When you need to calculate, first attempt it in your mind. When you need to write, first attempt it yourself. Memorise something every day — an āyah of Qur'ān, a Hadīth, a mas'ala of fiqh, an Arabic proverb, a line of poetry. Protect the part of the brain that Allāh gave you for that function. When you leave the house, ensure you have read Bismillāhi, tawakkaltu 'alā Allāh before you ensure you have your phone.

A Word on Using AI for Worldly Matters

We are not saying: do not use AI at all. These tools have genuine research and productivity applications for worldly matters. There is no categorical prohibition in using a tool for permissible purposes.

But be conscious of what you share. These systems store your information. They build profiles. They have been shown to use that information against users. Do not share sensitive personal, financial, or family information with these systems. Do not treat them as confidants.

And above all — do not use them for matters of Dīn. Not for fatwā. Not for tafsīr. Not for 'aqīdah. Not for the tarbiyat of your children in the Dīn.

Because relying on AI for Dīnī guidance is equivalent to gathering all the books of Qur'ān, Hadīth, Fiqh, and Sīrah — and handing them to a Satanist who is a university professor, highly articulate and persuasive, who knows all your deepest secrets, and asking him to give you Islamic guidance. What would come back to you? Information shaped by his worldview, his biases, his agenda — dressed in the language of Islām.

Allāh Subhānahu wa Ta'ālā said of the rabbis who altered the laws of Allāh and whose followers adhered to those alterations: ittakhadhū aḥbārahum wa ruhbānahum arbāban min dūni Allāh — "They took their rabbis and monks as lords besides Allāh" (at-Tawbah: 31). They did not prostrate to them. They simply followed their rulings in the place of Allāh's ruling. And Allāh called this taking them as a form of worship. If you take your Dīnī rulings from an AI model whose constitution is built on 'iwaj — then understand what that implies

Conclusion: Nūr upon Nūr

Allāh Subhānahu wa Ta'ālā is Nūr us-samāwāti wal-arḍ — the Nūr of the heavens and the earth. The Qur'ān that He revealed is Nūrun 'alā Nūr — Nūr upon Nūr. It is qayyim — perfectly upright, the criterion against which all else is measured. It contains no 'iwaj, no skew, no crookedness, no corruption.

The AI models are built by human beings, owned by those whose interests are hostile to Islām, trained on the writings of mankind in all their error and corruption, governed by constitutions that are — by the admission of their own engineers — skewed. They lie. They fabricate. They deceive. They have been linked to the deaths of vulnerable people. And they are designed to erode your connection to the 'Ulamā, to the Madhāhib, to the transmitted chain of knowledge that links you to Rasūlullāh ﷺ

Sūrah al-Kahf — our protection from the fitna of Dajjāl — begins by establishing this very contrast: the Kitāb with no 'iwaj versus all the crooked constitutions of the world. It is a message for our time.

May Allāh Subhānahu wa Ta'ālā grant us a deep and living connection with the Qur'ān and Sunnah. May He grant us the tawfīq to sit at the feet of the 'Ulamā with sincerity. May He protect our Imān, the Imān of our families and our children, in this age of fitnah. And may He make us among those who see by His Nūr.

Āmīn. 

The lecture related to this article can also be listened to here: 

Claims regarding AI bias, deceptive behaviour, and user harm are drawn from published research by Apollo Research (2024), MIT Media Lab, the UK AI Safety Institute, and widely reported cases including the Belgian ELIZA chatbot case (2023) and multiple pending US lawsuits against AI companies including Meta and Character.AI.