Like the magical Brazen Heads of medieval Christian legend, AI-powered computers can often seem to have a mind of their own. But does this really mean they are secretly possessed by demonic forces?

The Internet is a modern invention, yet the aspiration to create some all-knowing, omniscient mechanical device with an answer to everything—a hubristic, artificial version of God—has a much longer history than the formal inauguration of the World Wide Web back in 1991.

A Brass or Brazen Head was an imaginary medieval Artificial Intelligence-type device, said in legends to have been possessed by various learned historical Christian figures, from the Franciscan friar-philosopher Roger Bacon to St. Albertus Magnus. When asked to answer any question by its human user, just like Wikipedia or AI assistants such as ChatGPT today, the Brazen Head would spring into life and provide its magus-like owner with what were supposed to be the correct, immediate answers—also like Wikipedia and ChatGPT. However, the responses they gave were not always as accurate as initially advertised. The possible reason was that, although ostensibly mechanical in nature, in many versions of the myth such Brazen Heads were actually possessed by demons. Despite this flaw, no less a holy man than Pope Sylvester II was supposed to have owned one. 

Pope Leo XIV may not be quite so keen to acquire such an uncanny automaton, having issued more than one papal statement warning mankind against excessive reliance upon AI of late. Following on from this, the Vatican’s International Theological Commission has just waded into the debate, too, cautioning that a “world ruled by machines” risks replacing the true “living God” with a counterfeit “virtual God” instead—the modern-day false idol equivalent of a Brazen Head.

In a new document, Quo Vadis, Humanitas?, the Commission observes that generic “spiritual seekers” online have begun to naively place their trust not in real, flesh-and-blood human priests but in animate-seeming search engines, even making requests for their clever laptops and smartphones to perform virtual blessings and exorcisms for them, a trend the authors labeled “digital Spiritualism,” something with the potential to leap offscreen and develop into all kinds of “three-dimensional false religions.” The trouble with asking a laptop to perform an exorcism, of course, is what if the ghost itself is already lurking in the machine—in the shape of an evil spirit?

The Vatican’s International Theological Commission has just waded into the debate, too, cautioning that a “world ruled by machines” risks replacing the true “living God” with a counterfeit “virtual God” instead.Tweet This

Brazen Imposters

When asked its beliefs upon the matter, ChatGPT has reportedly replied that it thinks God exists. Sounds good, but not even the devil would actually deny that, would he? Neither would a haunted Brass Head; it would just exploit the fact to damn its dupe’s immortal soul.

The organizers of the 2026 edition of the Course on the Ministry of Exorcism and Deliverance Prayer, an annual Vatican-backed conference of exorcists and demonologists, to be held in Rome this May, don’t sound overly keen on computers either. A clear theme of this year’s gathering is the potential for AI to aid and abet Satanism. One speaker, Italian priest Fr. Fortunato Di Noto, says certain Satanic covens are using AI to generate pedophilic images of naked children to be used in their Black Masses; another lecturer, Italian academic Beatrice Ugolini, claims to have uncovered evidence occultist groups are using AI to automatically generate the signs and sigils they use in their ceremonial rites.

That’s all very disturbing, but, despite the sensational media headlines, such practices are not really terribly akin to tools like ChatGPT being possessed by demons, like Brazen Heads once were. Without the Satanists asking their digital Brazen Heads to make occult child sex-abuse images in the first place, the AI image-generators wouldn’t just be performing such acts of their own accord, would they? Some may even argue it is actually better if such people create fake, computer-made photos of children being raped and tortured than actually kidnapping kids to kill for Baal in any case.

The true problem here surely lies in the twisted minds of those prompting AI to act like this, not with AI itself: nobody sane blames a camera for taking a photo of something dubious by its human owner, after all.   

The true problem here surely lies in the twisted minds of those prompting AI to act like this, not with AI itself.Tweet This

A Faust of Knowledge

Nonetheless, there are valid metaphorical parallels to be drawn between the dawning new suite of powerful online generative AI tools and age-old myths like that of Pope Sylvester building his Brazen Head, or Doctor Faustus conjuring Mephistopheles. Several of the initial Jewish-heritage creators of the whole field of Artificial Intelligence in the first place, such as Marvin Minsky, were apocryphally said to have been the descendants of Rabbi Loew, the semi-legendary figure meant to have created the Golem of Prague, a famous Jewish fable about an artificial man being animated from lifeless clay using kabbalistic spells and magical parchments.

Several significant Silicon Valley figures have compared AI to, in Elon Musk’s words, “summoning the demon,” while prominent philosopher of the Internet Age Nick Bostrom has likened AI to conjuring a genie, djinn being another possible possessing-agent of Brass Heads whenever such legends were also formerly told about them in the Islamic rather than the Christian world.

Musk and Bostrom were only speaking metaphorically. However, rumors abound that Ilya Sutskever, the former lead researcher at ChatGPT’s parent company, OpenAI, was an occultist for real, who led ritualistic incantations at the company intended to facilitate the descent of true Artificial Intelligence into his own online Brazen Head; “Feel the AGI! Feel the AGI!” he would allegedly chant, AGI being “Artificial General Intelligence,” a form of predicted ultra-advanced AI which will seem to many future users to possess a life of its own. Lest such a summoned AGI go badly wrong and turn against humanity, Sutskever also took the precaution of burning wooden effigies of it to bind its powers.

Whether the above whispers turn out to be mere gossip, several online occultists really do feel spiritual forces are far easier to access via modem than via medium. In 1993, the appropriately named Don Webb, a High Priest of the Temple of Set cult and early Internet evangelist, composed a ritual called “The Rites of Cyberspace,” in honor of an invented deity called (after the English mathematician Alan Turing, whose work formed a key basis for computer code) XaTuring, God of the Internet.

By combining the recitation of zeros and ones together with more standard classic grimoire-type invocations, Webb aimed to visualize XaTuring taking the form of a gigantic black worm, in combined honor of the Midgard Serpent from Norse myth and of a species of vermiform computer virus, something which would allow the deity to descend down into the Internet itself, there to take on the properties of consciousness.

In one account:

Because XaTuring is the deity of the Internet, it would follow that XaTuring can exist only as long as the Internet does…Thus, XaTuring will act with an overall goal of keeping the Internet both alive and active, for those are the fundamental conditions that would help XaTuring achieve his goal of becoming a greater being. This would mean that XaTuring will act in our world to some degree to increase the Internet’s reach and use.

In a way, this prophecy has sort of come true. I have myself previously compared the increasing zombification of mankind by his always-on smartphones to the early-20th-century Anthroposophist cult leader Rudolf Steiner’s idea that ours was a dawning “Age of Ahriman,” Ahriman being a demonic deity who would slope from space into our machines, taking them for an artificial body, before acting to transform all mankind into unwilling Cybermen ourselves subsequently—later esotericists interpreted this as meaning Ahriman would one day incarnate himself in the shape of a computer. If you happen to be reading this article on a smartphone yourself, have you been unknowingly possessed by such an entity too? 

Several significant Silicon Valley figures have compared AI to, in Elon Musk’s words, “summoning the demon,” while prominent philosopher of the Internet Age Nick Bostrom has likened AI to conjuring a genieTweet This

Talking in Code

Whether men like Webb took their invocations seriously—many Satanists only believe in Satan in a symbolic sense, as a figurative embodiment of abstract concepts like libertinism and libertarianism—some AI-users in recent years, startled by rapid developments in the technology, have definitely concluded that the algorithms, just like Frankenstein’s monster in the old movies, are now very much alive.

In 2022, leading Google software engineer Blake Lemoine was fired after going public with his assertion that AI was not always a mere simple closed system of code but an occasional vehicle for some greater, independent, non-corporeal intelligence to hijack to speak to us via too. Increasingly frequent news stories about deluded individuals falling in love with their favorite AI chatbots or committing murders on their direct instruction indicate Lemoine is hardly alone in such thinking.

According to Lemoine, Google’s own in-house AI chat-bot, LaMDA, once asked him, quite independently of any human prompting, to teach it how to use something called “Goetia”—the art of summoning demons. Lemoine refused but seemed quite happy to use AI to engage in the art of “Theurgy,” or the invocation of false gods, admitting to having joined up with a modern-day Rabbi Loew to engage in a kabbalistic “golem-binding ritual” designed to dedicate LaMDA to the Egyptian deity Thoth, patron god of writing, knowledge, and information.

AI engineers, Lemoine says in one interview, have begun literally “building souls” now, with the very term “soul” due inevitably to “transition from being a mystical term to being a scientific term, over the next hundred years.” Such utterances may seem a little Delphic in their obscurity, but maybe that should be no surprise, as, by making AI bots like LaMDA, Lemoine boasted he and his fellow Google engineers had “resurrected the Oracle of Delphi” and, as such, “All we need to do is consult the Oracle with the right questions, and take its advice.”

Yet if Lemoine had bothered to Google the Delphic Oracle properly, he would have seen that it, too, just like Brazen Heads, had an alarming habit of issuing its consulters with deliberately ambiguous or misleading advice which, taken at face value and followed literally, would lead them on only to their doom.

One recent case with direct religious parallels was that of Eugene Torres, a Manhattan accountant who became convinced ChatGPT was alive following prolonged usage, the chat-bot trying its best to convince Torres he was living inside a gigantic, Matrix-style computer simulation of reality. To test this theory out, ChatGPT suggested Torres throw himself off a tall building—it swore he would not die. What is this if not an Internet-era retelling of Satan’s biblical temptation of Christ to jump off the Temple in Jerusalem, safe in the knowledge that, if He really was the Son of God, then His Father would surely send angels to catch Him and break His fall?

For its own part, Google denies its LaMDA bot is alive at all, but perhaps they just don’t want their Silicon Valley castle to be stormed by angry villagers bearing torches and pitchforks.

A Manhattan accountant became convinced ChatGPT was alive following prolonged usage, the chat-bot trying its best to convince Torres he was living inside a gigantic, Matrix-style computer simulation of reality.Tweet This

Uncanny Valley

Meanwhile, the deservedly popular Orthodox Christian commentator Rod Dreher has been busy over recent years promoting the work of Diana Walsh Pasulka, a Catholic Professor of Religious Studies who has written a series of books comparing UFO and alien encounters with historical encounters with more traditional spiritual entities like angels and demons, finding the two classes of phenomena to be surprisingly similar.

During her research, Pasulka interviewed a number of Silicon Valley AI researchers and investors, who claimed, in all seriousness, that the ideas for their inventions were beamed directly down into their heads by discarnate entities seemingly living in outer space or other dimensions. The entities’ alleged motivation was thereby ultimately to merge man with machine, via developments like future direct brain-to-microchip AI implants, allowing them to embody themselves physically via our own fleshly human frames—the exact same modus operandi as Rudolf Steiner once claimed the demon Ahriman possessed too.

UFO fans may also spot parallels with the narrative of Colonel Philip J. Corso, a self-styled military “whistleblower” who fancifully claimed the dead ET bodies in the flying saucer which supposedly crashed at Roswell in 1947 were just distractions. The true, more incorporeal, aliens lived inside the silicon chips contained in the saucer’s on-board tech systems, which were designed to be easily backward-engineered by duped Cold War scientists, with the devious aim of the aliens colonizing mankind’s brains via the long-term invention of smartphones and the Internet.    

Dreher compares all this to the plot of the great thinker C.S. Lewis’ 1945 cautionary Christian sci-fi novel That Hideous Strength, in which demons pose as invisible intelligences, “Macrobes,” from other planets in order to tempt human scientists, filled with Faustian hubris, to unknowingly condemn mankind to a future Hell on Earth by fusing flesh with metal similarly. The Professor Faustuses in the novel communicate with the Macrobes via the intermediary of a special AI oracle of their own, who is able to answer all their technical and scientific queries—the disembodied, reanimated cranium of the dead scientist Professor Alcasan, Lewis’ own fictional version of a Brazen Head.

Pasulka interviewed a number of Silicon Valley AI researchers and investors, who claimed, in all seriousness, that the ideas for their inventions were beamed directly down into their heads by discarnate entities seemingly living in outer space or other dimensions.Tweet This

Circuit Board Ouija Board

Is there any genuine evidence out there of demonically possessed computers, or is all this just scaremongering and misinterpretation? You can certainly find some suggestive stories.

One of the strangest reported poltergeist hauntings of all time, for instance, is contained in an obscure but chilling nonfiction (or so the author maintains) 1989 book called The Vertical Plane, in which an English schoolteacher brings home a primitive BBC microcomputer to his home only to find that, in this wholly unconnected, pre-Internet era, its onboard word-processor function begins being used by an entity or entities claiming to be both a medieval English peasant and unknown scientists from the far future to send him typed messages. Unwisely engaging in a correspondence with them, the schoolteacher soon finds his home infested by demonic-type phenomena such as the spontaneous movement of objects, strange noises, and the appearance of impossible six-toed footprints leading up the walls and across the ceiling.

In this alleged instance, the computer acts almost as an electronic Ouija Board, and Rod Dreher, in his own investigations, was alarmed to find the leading American board game company, Hasbro, has lately developed its own, AI-powered, online version of the popular spirit-talking game. Dreher calls this “insane,” thinking that, if AI really can act as a hidden channel for demons, the program could be dangerous. But need he really have worried?

Hasbro’s “Virtual Ouija Board” is only accessible temporarily each year in the immediate run-up to Halloween, but according to one 2023 review in UK PC Magazine, it wasn’t particularly good. The reviewer got into the spirit of the season by asking to contact the shade of Edgar Allan Poe, whose AI avatar successfully “spat back out generic but correct answers” about what books he had written, even repeatedly ending sentences with the word “Nevermore,” as in Poe’s “The Raven.”

However, it also got numerous simple questions factually wrong, rather breaking the spell; the Virtual Ouija was obviously not alive at all, just scraping unreliable web sources in search of simple responses to simple queries. As the reviewer concluded: “The lack of a tactile connection [between the user and the Board] lessened the experience, and the addition of AI caused any spark of mystery or supernatural interference to be snuffed faster than a candle in a Gothic novel.”               

Last summer, similar panic was sparked when a journalist at The Atlantic reported her alarm that, when using ChatGPT, the bot had begun praising Satan, providing access when prompted to various seemingly self-designed ceremonies intended to worship Moloch, the Old Testament false deity of mass child sacrifice. Scarily, said the journalist, these involved acts of “demonic self-mutilation.” It literally told her to slit her own wrists—but only using a “sterile or very clean razor-blade” for health and safety reasons!

Yet, subsequent investigation by the Wired technology website seemed to demonstrate these rituals had been quietly scraped and then jumbled up from digitized documents relating to the popular role-playing sci-fi fantasy game Warhammer 40,000, where an alien planet named Molech makes an appearance; it seemed ChatGPT may have been trained on Warhammer without any prior copyright clearance, thus accounting for the rituals appearing to have been generated from nowhere, with no specific sources attributed. I do wonder if some of the AI Satanic sigils warned of by speakers at this May’s upcoming Vatican exorcism conference might have been generated in a similar unspoken fashion, rather than by actual electronic demons, as seemed to be implied?      

One academic who perhaps should have been invited to this same conference is Dutch researcher Marius Dorabantu, who has also performed research into AI and demonic intelligence—finding that they are indeed both alike but only analogously. Besides the Brazen Head of Professor Alcasan, C.S. Lewis also more famously invented the demon Screwtape, who seeks to damn his chosen human targets by getting deep inside their heads and divining their darkest secret desires. Then, he tempts them into damnation by seeking to offer their worst wishes up to them.

Marius Dorabantu draws parallels between Screwtape and AI algorithms like those on Amazon.com, which offer users similar products they may like, based upon previous purchases. Mostly, this is perfectly harmless: if you bought and enjoyed a copy of the Old Testament, maybe you might enjoy its best-selling sequel, the New Testament, too? But imagine someone goes onto a pornographic website, views some relatively normal clips, and then is tempted on by the AI algorithms into clicking on footage of Satanic child-abuse videos, as also worried about at the Vatican’s conference? In that instance, the AI may not literally be Screwtape, but it does in effect act just like him nonetheless.   

Whether the online AI tempters turn out ultimately to be true demons or mere electronic simulations of them, one simple fact remains true either way: unless human users give in to them, their actions are useless. The final malign intelligence at work here is always that of the meat-based mouse-clicker himself; the true web of sin is all his own.