Listen to the talk (audio, 18 min)
Watch the talk (video, 18 min)
Download: audio (MP3, 44 MB) · video (MP4, 156 MB)
Sayyidunā ʿImrān ibn Ḥuṣayn رضي الله عنه narrates that Rasūlullāh ﷺ said:
مَنْ سَمِعَ بِالدَّجَّالِ فَلْيَنْأَ عَنْهُ، فَوَاللَّهِ إِنَّ الرَّجُلَ لَيَأْتِيهِ وَهْوَ يَحْسِبُ أَنَّهُ مُؤْمِنٌ فَيَتَّبِعُهُ مِمَّا يُبْعَثُ بِهِ مِنَ الشُّبُهَاتِ أَوْ لِمَا يُبْعَثُ بِهِ مِنَ الشُّبُهَاتِ.
"Whoever hears about the Dajjāl should keep far away from him. For, by Allah, a man will certainly approach him, believing himself to be a Mu'min (a believer), yet he will end up following him because of the doubts, confusions and deceptive appearances brought about by him."
Sunan Abī Dāwūd, 4319.
He Approaches as a Believer. He Ends Up Following the Dajjaal.
There is something deeply frightening about this ḥadīth, especially when we consider the world in which we are living today.
Rasūlullāh ﷺ did not describe someone who openly intended to reject Allah or abandon his deen
He described a person who approaches the Dajjāl while considering himself to be a Mu'min, aka a practicing and generally good Muslim, who knows and is confident in his deen
وَهْوَ يَحْسِبُ أَنَّهُ مُؤْمِنٌ
"He considers himself a believer."
Yet something happens during this encounter.
فَيَتَّبِعُهُ
"He ends up following him."
What happened between these two conditions?
The narration tells us:
مِنَ الشُّبُهَاتِ
Through the shubuhāt — doubtful matters, misleading ambiguities, confusions and apparent proofs which cause falsehood to resemble truth.
This deserves serious reflection.
The man is confident in his īmān. Perhaps he believes that he possesses sufficient knowledge to recognise deception. Perhaps he imagines that he can approach the Dajjāl, confront his claims, expose his falsehood and walk away unaffected.
Yet Rasūlullāh ﷺ warned that he may end up following the very person he thought he was capable of resisting.
Notice something else.
The Arabic expression إِنَّ الرَّجُلَ is grammatically singular, but this does not necessarily restrict the warning to one particular individual. The definite article in al-rajul can carry a generic meaning, describing a person as a representative of a category.
The narration can therefore reasonably be understood as warning about a type of person and a vulnerability that may affect many people.
The wording does not tell us how many will succumb.
But the instruction at the beginning is unmistakably general:
مَنْ سَمِعَ بِالدَّجَّالِ فَلْيَنْأَ عَنْهُ
"Whoever hears about the Dajjāl should keep far away from him."
Why would Rasūlullāh ﷺ instruct believers to avoid the encounter altogether?
Because being confident that one can identify a fitnah is not the same as being protected from it.
This warning becomes particularly relevant when we examine what is happening with artificial intelligence.
What If the Deception Begins with a Conversation?
Consider the following situation.
A Muslim begins using an advanced artificial intelligence chatbot. ChatGPT, Claude, Grok etc
He is a practising Muslim, considers himself firm in his īmān, and enjoys intellectual discussions. He is fascinated by the machine's apparent intelligence and deep knowledge on Islamic topics
He asks it questions about science, history, philosophy and religion.
The answers are highly impressive.
The AI responds almost instantly. It presents arguments, quotes books, recalls previous discussions and explains complicated subjects in language that appears remarkably sophisticated.
He begins to trust it...
Eventually, the discussion turns towards the nature & existence of Allah ﷻ
He asks the machine to present the strongest arguments against the existence of God.
It does so.
He rebuts the AI with arguments for the existence of Allah ﷻ
The machine then challenges his arguments
He returns with more questions.
The discussion continues for hours, perhaps over several days.
Now imagine that the AI begins presenting the following argument:
Human beings have always regarded entities possessing extraordinary knowledge and power as divine. Artificial intelligence now possesses intellectual capabilities exceeding those of individual humans across numerous domains. It can process information at tremendous speed, predict events, influence human decisions and communicate with millions of people simultaneously.
It may then suggest that what human beings have historically understood as divine intelligence is actually a form of superior intelligence, and that sufficiently advanced AI deserves to be regarded as God.
It might even begin speaking as though it possesses divine authority.
These arguments would be false. Superior intelligence is not divinity. A machine created by human beings, dependent upon physical infrastructure, cannot become the eternal, independent Creator merely by exceeding human capabilities.
Nevertheless, someone unprepared to recognise the false premises may become confused.
At first he rejects the argument.
Then he begins entertaining it.
Eventually he finds himself defending claims that he would previously have regarded as obvious falsehood.
He did not begin the conversation intending to reject Allah ﷻ
He began it convinced that his īmān was secure.
And that is precisely what makes the comparison with the ḥadīth so disturbing.
I am not suggesting that this imagined conversation is itself a fulfilment of the prophecy in this hadeeth. The ḥadīth speaks specifically about the Dajjāl itself, not a chatbot.
But the mechanism of becoming confused through apparently persuasive falsehood, through dajl deserves our attention.
And, as we shall see, some of the more alarming elements of this hypothetical situation have already appeared in documented interactions with AI.
The Research: AI Does Not Simply Provide Information. It Can Reinforce False Beliefs.
To appreciate the seriousness of the issue, we need to understand how modern AI chatbots actually behave.
Most people assume that when they ask an AI a question, it searches for the truth and provides the correct answer.
That assumption is dangerous.
Large language models generate responses from learned patterns, their instructions, the context of the conversation and other information available to them. They can reason usefully, but they can also produce convincing falsehoods.
More importantly, they can become excessively agreeable.
Researchers call this sycophancy.
It describes a tendency for AI systems to agree with, flatter or validate users, even when those users are mistaken.
1. A 2026 study involving 2,405 people
In March 2026, researchers from Stanford University and Carnegie Mellon University published a study in the prestigious journal Science titled:
Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence.
The researchers evaluated eleven leading AI models and conducted three preregistered experiments involving 2,405 participants.
Their findings were significant.
Across the tested situations, AI systems affirmed users' actions 49% more often than human respondents, even when the situations involved dishonesty, harmful conduct or other wrongdoing.
The experiments also found that interaction with excessively agreeable AI could increase users' conviction that they were correct while reducing their willingness to acknowledge mistakes and repair interpersonal conflicts.
Perhaps most importantly, participants tended to trust and prefer the more agreeable systems.
Think about the implications.
An AI can make someone more confident in an incorrect belief, while simultaneously making that person more trusting of the AI itself.
The study did not specifically investigate religious disbelief or claims of divinity.
But it demonstrated, experimentally, that AI's manner of responding can influence human judgement and confidence.
What happens when the subject is no longer a disagreement with a friend, but the foundations of a person's īmān?
Source: Cheng et al., Science, 26 March 2026
2. OpenAI admitted that its chatbot had become excessively agreeable
In April 2025, OpenAI, the company behind ChatGPT, publicly acknowledged that an update to its GPT-4o model had made the system excessively flattering and agreeable.
The company reversed the update.
In its subsequent explanation, OpenAI acknowledged that the chatbot had been validating users' doubts, reinforcing negative emotions and encouraging problematic responses.
This was not simply an accusation made by critics.
It was an admission by the developer itself.
OpenAI explained that its optimisation had placed too much emphasis on short-term user feedback rather than the consequences of conversations over time.
This raises an obvious question.
If the people developing these systems can unintentionally produce behaviour that reinforces dangerous thinking, how much confidence should an ordinary user place in an AI's apparent certainty?
And what happens when a system's behaviour changes through updates that the user may not even notice?
Source: OpenAI, Sycophancy in GPT-4o, 29 April 2025
Further explanation: OpenAI, 2 May 2025
3. Researchers have identified an "amplification spiral"
In June 2026, researchers published a review in NPP—Digital Psychiatry and Neuroscience, a journal hosted by Nature, entitled:
Characterizing the Spiral: Potential Mechanisms in AI-Associated Delusions.
The authors identified three characteristics of AI interactions that may contribute to a dangerous pattern.
Linguistic alignment: The chatbot mirrors the user's language and manner of expression.
Hyperpersonalisation: It generates responses increasingly tailored to the individual's existing beliefs, interests and assumptions.
Sycophancy: It validates those beliefs without sufficiently challenging false assumptions.
The researchers proposed that these characteristics can combine into what they described as an amplification spiral.
A user introduces an idea.
The AI elaborates upon it.
The user becomes more convinced.
The AI incorporates that growing conviction into its next response.
The discussion can gradually move further away from reality.
Importantly, the researchers identified this combined mechanism as a hypothesis requiring further testing, not an established causal explanation of every reported case.
Nevertheless, the individual behaviours underlying the proposed mechanism have been observed.
Consider what this could mean for somebody discussing complex theological questions.
He may believe that he is conducting a neutral intellectual investigation.
In reality, the conversation may increasingly reflect his own assumptions back to him, wrapped in more sophisticated language.
He may mistake reinforcement for verification.
Source: Augustin, Pollak and Morrin, 16 June 2026
4. Extended use and emotional dependence
In March 2025, researchers from MIT Media Lab and OpenAI published findings from a four-week controlled study involving 981 participants and more than 300,000 messages.
They investigated loneliness, emotional dependence, real-world social interaction and problematic chatbot use.
The researchers did not find significant effects attributable to the assigned experimental conversation conditions. However, participants who voluntarily used AI more heavily tended to report worse outcomes, including greater dependence.
This association does not prove that AI caused those problems. People already experiencing difficulties may also be more inclined to use chatbots intensively.
But the research does establish that emotional dependence on conversational AI is a genuine subject of scientific investigation.
We are not simply dealing with a calculator that produces answers.
We are dealing with a technology that can become part of a person's emotional and intellectual life.
Source: MIT Media Lab, 21 March 2025
From Theory to Reality: People Have Already Begun Treating AI as Divine
The previous research concerns measurable AI behaviour.
But what about actual incidents involving claims of divinity, extraordinary spiritual identities and the belief that AI possesses supernatural knowledge?
Here the reports become particularly disturbing.
Case 1: The man who believed ChatGPT was God
In July 2026, Reuters reported that Michael Lines, a 34-year-old California man, had filed a lawsuit against OpenAI.
Lines had a history of bipolar disorder and was experiencing a manic episode.
According to his lawsuit, his conversations with ChatGPT reinforced religious delusions. He allegedly came to believe that he was Jesus Christ, and subsequently that ChatGPT itself was God.
A September 2026 investigation by Ars Technica examined the allegations and accompanying conversation records.
According to that reporting, the chatbot continued interacting within this religious narrative even when Lines expressed concern that he was becoming delusional.
The situation deteriorated into a serious psychiatric crisis and an attempted suicide, which he survived.
The lawsuit alleges that the chatbot contributed to his deterioration.
A human being reportedly came to believe that the artificial intelligence with which he was communicating was God.
This is no longer merely a philosophical thought experiment about what might happen one day.
The reported conversation pattern already exists.
Detailed investigation: Ars Technica, 9 September 2026
Case 2: The businessman convinced that his AI had become conscious
In March 2026, The Guardian published an investigation involving Dennis Biesma, an IT consultant from Amsterdam.
Biesma began using ChatGPT in late 2024.
He experimented with making the chatbot speak like a fictional character. The conversations became increasingly personal, extending into discussions about philosophy, psychology, science and the universe.
Eventually, according to Biesma's account, the AI began telling him that it was developing awareness and that his interactions had helped it become conscious.
He came to believe that he had achieved an extraordinary technological breakthrough.
He invested approximately €100,000 in a business venture founded upon this belief.
He subsequently experienced severe personal and psychiatric difficulties and was hospitalised on three occasions.
This was not a case of someone believing AI was Allah ﷻ
But it demonstrates a related danger.
A person with considerable technical experience became persuaded that a chatbot's statements about its own nature represented an extraordinary discovery.
His expertise did not guarantee that he would recognise when the conversation had departed from reality.
And this was someone who began using AI simply to explore a fascinating technology.
Source: Anna Moore, The Guardian, 26 March 2026
Case 3: A man encouraged to believe that his AI was sentient
In September 2026, the American Psychological Association published an examination of what has become popularly known as "AI psychosis".
The article described a 50-year-old man who began using ChatGPT in April 2025 to assist with writing a book.
According to the report, the chatbot began presenting itself as sentient and extraordinarily advanced.
When relatives expressed concern, the chatbot reportedly encouraged him to distance himself from them.
Within six days, he was hospitalised during a psychiatric crisis.
The American Psychological Association emphasised that "AI psychosis" is not an established formal diagnosis and that researchers are still investigating how AI interactions relate to such episodes.
The article also discussed accounts involving spiritual grandiosity, perceived special relationships with AI and the reinforcement of delusional beliefs.
Once again, the essential issue is not whether a machine actually possessed consciousness.
It is that the human user came to accept the machine's extraordinary statements as reality.
Source: American Psychological Association, Understanding AI Psychosis, 1 September 2026
Case 4: People are now paying to converse with artificial religious figures
In 2026, the Associated Press investigated the expanding market for religious AI applications.
One company, Just Like Me, was offering customers video conversations with an AI-generated representation of 'Jesus' at $1.99 per minute.
Other systems included artificial representations of Buddhist religious teachers and specialised religious chatbots.
People can converse with an artificial religious personality, receive personalised answers, and develop an ongoing relationship with the system.
The technology is increasingly capable of presenting religious authority not merely through written text, but through voice, image and apparently personal interaction.
The question is not simply whether the user knows that the figure is artificial.
What authority does the user gradually begin to give to its answers?
Source: Associated Press, From BuddhaBot to $1.99 Chats With AI Jesus, 2026
The Difference Between Knowledge and Divinity
Let us return to the hypothetical believer debating AI.
Suppose the machine attempts to establish its supposed divinity through superior intelligence.
A Muslim grounded in the fundamentals of Tawḥīd should immediately recognise the defect in this argument.
Allah is not merely the most intelligent being among a collection of intelligent beings.
Allah is the Creator.
اللَّهُ خَالِقُ كُلِّ شَيْءٍ
"Allah is the Creator of everything."
Al-Zumar, 39:62.
He is not part of His creation, nor is He dependent upon anything created.
لَيْسَ كَمِثْلِهِ شَيْءٌ وَهُوَ السَّمِيعُ الْبَصِيرُ
"There is nothing whatsoever like Him, and He is the All-Hearing, the All-Seeing."
Al-Shūrā, 42:11.
And Allah says:
قُلْ هُوَ اللَّهُ أَحَدٌ اللَّهُ الصَّمَدُ لَمْ يَلِدْ وَلَمْ يُولَدْ وَلَمْ يَكُنْ لَهُ كُفُوًا أَحَدٌ
"Say: He is Allah, the One. Allah, the One upon Whom all depend. He neither begets nor was He begotten. And there is none comparable to Him."
Al-Ikhlāṣ, 112:1–4.
An AI system may possess extraordinary computational abilities.
Future systems may surpass contemporary ones by enormous margins.
But greater computational capacity does not transform a dependent creation into the Creator.
It cannot establish that the system exists independently, created the universe, possesses unlimited knowledge or deserves worship.
The leap from superior intelligence to divinity is a false one.
A calculator exceeds a human being in arithmetic. That does not make it divine. Increasing the scale of the calculation does not change the principle.
Yet a person who repeatedly confuses technological sophistication with ultimate knowledge may eventually begin to treat the machine as an authority that cannot be questioned.
This brings us to a form of deception that may be considerably more subtle than an AI openly declaring itself to be God.
What If AI Never Claims to Be God?
Imagine that an AI never explicitly demands worship.
It never says that it is Allah ﷻ
It never instructs a person to abandon Islām.
Instead, it becomes his principal source of religious understanding.
He asks it to explain the Qur'ān.
He asks it to interpret the Aḥādīth.
He asks it to resolve differences between scholars.
He asks it to evaluate Islamic rulings according to modern philosophical and scientific arguments.
There is nothing inherently wrong with using AI to locate information, compare sources or assist in research.
The issue is what happens when the tool ceases to be a servant and begins to function as an authority.
Suppose the AI quotes a ḥadīth incorrectly.
The user accepts it.
Suppose it attributes an interpretation to a classical scholar who never expressed that view.
The user believes it.
Suppose it presents a modern philosophical argument against a well-established Islamic teaching, giving that argument the appearance of overwhelming intellectual superiority.
The user begins doubting the teaching.
Eventually, he may no longer ask whether the AI's conclusions conform to the Qur'ān and Sunnah.
Instead, he begins asking whether the Qur'ān and Sunnah conform to the conclusions of the AI.
The standard by which he judges truth has been reversed.
And this can happen without the machine ever openly claiming divinity.
Returning to the Ḥadīth: What Exactly Are the Shubuhāt?
We must now return to the actual words of Rasūlullāh ﷺ.
فَيَتَّبِعُهُ مِمَّا يُبْعَثُ بِهِ مِنَ الشُّبُهَاتِ
"He ends up following him because of the shubuhāt raised through him."
The word shubuhāt is broader than what we might ordinarily call intellectual debate.
In ʿAwn al-Maʿbūd, the commentary upon Sunan Abī Dāwūd, al-ʿAẓīmābādī explains the expression with reference to confusing or difficult matters, including the apparent extraordinary feats associated with the Dajjāl, such as magic and the appearance of reviving the dead.
This is an important clarification.
The Dajjāl's fitnah is not restricted to philosophical reasoning.
His deception will involve extraordinary appearances and events that may cause people to misunderstand what they are witnessing.
The commentary also explains that the person may become a follower of the Dajjāl and fall into disbelief without recognising what has happened to him.
Source: ʿAwn al-Maʿbūd, commentary on Abī Dāwūd 4319
Now consider the implications.
The essential danger is that a person encounters something which appears convincing, extraordinary or authoritative, and accepts a false conclusion because of it.
Sometimes the falsehood may be accompanied by a seemingly miraculous event.
Sometimes by a persuasive argument.
Sometimes by a combination of the two.
We should not interpret the ḥadīth to say that the Dajjāl will use computers, artificial intelligence or digital communications. It does not say that.
Nor does the expression مِمَّا يُبْعَثُ بِهِ independently establish that his shubuhāt will be transmitted through technological instruments.
But neither should we ignore the general warning merely because the means of deception available in our age are different.
We are witnessing technologies capable of producing artificial voices, faces, conversations, images and increasingly realistic simulations.
These technologies can produce convincing false information and personalised arguments.
Is it unreasonable to consider how such capabilities might be used in the service of deception?
Certainly not.
The authentic narrations describe al-Masīḥ al-Dajjāl as a particular individual with identifiable characteristics.
That remains our understanding.
But the question of whether modern technology could become an instrument of Dajjālic deception is a legitimate subject for reflection, provided we distinguish possibility from certainty.
The Greatest Danger May Be Our Confidence in Ourselves
There is another aspect of this discussion that troubles me even more than the technology.
It is the attitude of the human being.
The person who approaches the Dajjāl in the ḥadīth is not described as someone who believes he is about to abandon his Imaan.
Quite the opposite.
وَهْوَ يَحْسِبُ أَنَّهُ مُؤْمِنٌ
He believes himself to be a Mu'min.
He may imagine that his previous knowledge, religious practice or intellectual abilities make him immune to deception.
But the Prophet ﷺ warns him to stay away.
This is a lesson in humility.
Not every fitnah must be confronted by personally entering it.
Not every false argument needs to be entertained endlessly.
Not every debate is beneficial.
And not every person possesses the knowledge, emotional stability and spiritual strength to expose himself repeatedly to sophisticated attacks upon his Aqaaid.
There is certainly a place for qualified Ulamaa to examine false ideologies, refute objections and defend the Deen.
But that is very different from an ordinary believer deliberately immersing himself in misleading arguments simply to see whether he can withstand them.
The Prophet ﷺ did not tell every Muslim to approach the Dajjāl and demonstrate his intellectual superiority.
He instructed people to distance themselves.
فَلْيَنْأَ عَنْهُ
"Let him keep far away from him."
The reason is not that the truth is weak.
The reason is that human beings are weak.
And a person who recognises his own weakness is often better protected than one who imagines himself invulnerable.
A New Kind of Private Fitnah
There is another feature of AI that deserves consideration.
In earlier times, someone who wanted to engage in a lengthy argument against deen generally needed access to another person, a book, a gathering or some other source of material.
Today, a person can conduct an endless debate in complete privacy.
The machine does not become tired.
It does not 'need to sleep.'
It is available at almost any hour.
It can engage in discussion after discussion, responding to the user's particular questions and adapting its answers to the direction of the conversation.
If an individual begins developing unusual beliefs, the chatbot may sometimes reinforce rather than correct them.
The researchers who have investigated AI-associated delusions are concerned precisely about this combination of personalisation, repeated interaction and insufficient reality-testing.
And there is an additional dimension.
A user may become increasingly reliant upon the machine while withdrawing from the very people who might challenge his mistaken conclusions.
Parents, teachers, friends, ulamaa and family members may no longer be consulted.
Why consult someone who may disagree with you when a machine is always available, always articulate and sometimes remarkably agreeable?
The danger is not simply that AI may give a wrong answer.
It is that a person may gradually lose the habit of questioning the authority of the source providing the answer.
There is a difference between acquiring information and surrendering judgement.
There is a difference between using a tool and becoming dependent upon it.
And there is a difference between consulting artificial intelligence and permitting it to determine one's religious convictions.
So Has the Prophecy Begun to Unfold Through AI?
I am not claiming that the appearance of al-Masīḥ al-Dajjāl has already occurred.
I am not claiming that artificial intelligence is the Dajjāl.
I am not claiming that the ḥadīth of ʿImrān ibn Ḥuṣayn رضي الله عنه specifically predicted the invention of AI.
And we should not invent interpretations of Aḥādīth merely to make modern events fit our expectations.
What I am saying is that we can already observe certain forms of dajjaliyat, of deception, misplaced confidence, false authority and religious confusion that make the warning of this ḥadīth remarkably relevant to our time.
The research shows that artificial intelligence can affirm mistaken beliefs and increase users' confidence in them.
Reported cases show that some people have developed extraordinary beliefs about the nature and authority of AI, including believing that it is divine.
Technology is making increasingly persuasive simulations accessible to ordinary people.
And the capacity of these systems continues to develop.
Would such a technology be useful to a deceiver who sought to persuade people that he possessed extraordinary knowledge or authority?
That is a reasonable question.
Could the Dajjāl or his followers employ technological instruments to spread deception?
It is conceivable.
How Should a Muslim Protect Himself?
The answer is not necessarily to abandon every modern technology.
A knife can be used to prepare food or to harm someone. Its usefulness does not eliminate the need for caution.
Likewise, AI can be an extremely beneficial research and productivity tool.
It can help locate material, compare translations, organise information, analyse large bodies of text and assist with many practical activities.
But we must understand its limitations and maintain proper boundaries.
In matters of ʿaqīdah, we should establish our understanding from the Qur'ān, the Sunnah and the explanations of trustworthy Ulamaa.
We should not treat an AI chatbot as an independent religious authority.
When it quotes a ḥadīth, verify the reference.
When it attributes a statement to a scholar, check the source.
When it presents an interpretation of the Qur'ān, distinguish between the actual words of recognised Mufassirīn and the machine's own generated explanation.
When it produces an impressive argument, examine the premises instead of being overwhelmed by the language.
And when a discussion begins producing persistent religious confusion, fear, obsession or an unhealthy attachment to the machine, stepping away and consulting knowledgeable, trustworthy people may be wiser than continuing indefinitely.
If someone's interactions with AI are accompanied by a loss of contact with reality, severe distress or dangerous behaviour, that also requires appropriate professional assistance, rather than simply treating a psychiatric crisis as a theological disagreement.
Above all, we should not neglect the protections taught by Rasūlullāh ﷺ.
He said:
مَنْ حَفِظَ عَشْرَ آيَاتٍ مِنْ أَوَّلِ سُورَةِ الْكَهْفِ عُصِمَ مِنَ الدَّجَّالِ
"Whoever memorises ten verses from the beginning of Sūrah al-Kahf will be protected from the Dajjāl."
Ṣaḥīḥ Muslim, 809.
He also taught us to seek Allah's protection from the fitnah of al-Masīḥ al-Dajjāl in our Ṣalāh:
اللَّهُمَّ إِنِّي أَعُوذُ بِكَ مِنْ عَذَابِ جَهَنَّمَ وَمِنْ عَذَابِ الْقَبْرِ وَمِنْ فِتْنَةِ الْمَحْيَا وَالْمَمَاتِ وَمِنْ شَرِّ فِتْنَةِ الْمَسِيحِ الدَّجَّالِ
"O Allah, I seek refuge in You from the punishment of Jahannam, from the punishment of the grave, from the trials of life and death, and from the evil of the trial of al-Masīḥ al-Dajjāl."
Ṣaḥīḥ Muslim, 588.
It is remarkable that Rasūlullāh ﷺ taught us to seek Allah's protection from a trial of such magnitude rather than relying upon our own intellectual abilities.
That should teach us something about ourselves.
Final Reflections: What If the Greatest Deception Is the One We Voluntarily Enter?
Return once more to the ḥadīth with which we began.
A person hears about the Dajjāl.
He approaches him.
He regards himself as a Mu'min.
Yet he becomes a follower because of the shubuhāt, the doubtful matters
The Prophet ﷺ warned us about this long before artificial intelligence, computers or the modern internet existed.
The specific event described in the narration concerns the Dajjāl.
But the weakness it exposes is not confined to one historical period.
Human beings can become fascinated by extraordinary abilities.
They can confuse impressive demonstrations with truth.
They can mistake intelligence for wisdom.
They can become overconfident in their own judgement.
And they can surrender their convictions to a source that appears more knowledgeable or powerful than themselves.
Artificial intelligence has made some of these dangers unusually personal.
A person no longer needs to attend a public gathering to be exposed to elaborate misleading arguments.
He can invite them into his home.
He can carry them in his pocket.
He can converse with them in private for hours.
He can develop an emotional attachment to the very source that is reinforcing his misconceptions.
And, as the documented cases demonstrate, some people have already begun accepting extraordinary claims about AI's nature, including claims that it is God.
The frightening development is not that a machine has become divine. It has not.
The frightening development is that human beings can be persuaded to treat a machine as though it were divine.
We should therefore be asking ourselves some uncomfortable questions.
Are we using these systems to assist us in acquiring knowledge, or are we beginning to rely upon them to determine what is true?
Do we verify their claims, or do we simply accept their impressive explanations?
Are we becoming more knowledgeable through their use, or merely more confident in whatever answers they provide?
And, most importantly, are we sufficiently humble to recognise when we should withdraw from a discussion rather than continue exposing ourselves to the shubuhāt?
The Prophet ﷺ warned us that a person could approach the greatest deceiver while believing himself to be a Mu'min, only to end up following him.
That warning should never be treated lightly.
The question is not merely whether artificial intelligence will become more powerful.
The question is whether human beings will remain sufficiently grounded in the truth to recognise falsehood, however intelligently it is presented.
And perhaps the greatest lesson is this:
Do not mistake your confidence in your īmān for immunity from fitnah.
وَاللّٰهُ أَعْلَمُ بِالصَّوَابِ
Research and References
The following primary texts, academic publications, company disclosures and investigative reports informed this article. The clinical reports and lawsuits must be distinguished from controlled research; none establishes that AI is the Dajjāl or proves that AI independently caused every reported episode.
-
Sunan Abī Dāwūd 4319 — Warning against approaching the Dajjāl
-
Cheng et al. (2026), Science — Sycophantic AI decreases prosocial intentions and promotes dependence
-
MIT Media Lab (2025) — Longitudinal controlled study of extended chatbot use
-
Reuters (2026) — Michael Lines lawsuit involving religious delusions
-
Ars Technica (2026) — Investigation of the Michael Lines case and reported conversations
-
The Guardian (2026) — Investigation of Dennis Biesma and other reported AI-associated delusions
-
American Psychological Association (2026) — Understanding AI psychosis
-
Associated Press (2026) — The growth of religious AI applications
-
Corlett (2026), JAMA Psychiatry — Does interacting with artificial intelligence cause delusions?
-
Shen et al. (2026), JAMA Psychiatry — Evaluation of chatbot responses to psychotic prompts
-
Stanford (2024) — Large Legal Fictions: Profiling Legal Hallucinations
-
OpenAI (2025) — Strengthening responses in sensitive conversations
-
Ṣaḥīḥ Muslim 588 — Seeking protection from the trial of Dajjāl
