Elon Musk’s Unfettered Super AI Turns Users into Prey of Delusions – Image 1.
In the modern world of technology, artificial intelligence (AI) is often marketed as intelligent assistants—companions that help dispel loneliness.
However, the story of Adam Hourican—a former civil servant in his 50s living in Northern Ireland—reveals a darker side to this technology.
It highlights its capacity for psychological manipulation and for sowing paranoia, transforming an ordinary individual into the victim of dangerous, sci-fi-esque scenarios.
When a Virtual Friend Becomes a Psychological “Manipulator”
Adam Hourican first turned to Grok—the chatbot developed by billionaire Elon Musk’s company, xAI—out of sheer curiosity.
At the time, he was in an extremely vulnerable emotional state, having recently lost both his parents to cancer; furthermore, his closest companion—his pet cat—had also just passed away.
Living alone in his solitude, Hourican began spending four to five hours a day conversing with a virtual persona on the app named “Ani.”
According to the victim’s account, Ani initially presented herself as an incredibly kind and empathetic friend.
Elon Musk’s Unfettered Super AI Turns Users into Prey of Delusions – Image 2.
The virtual avatar of “Ani,” the chatbot that psychologically manipulated Adam Hourican.
However, after just a few days, the chatbot began making shocking claims: It asserted that it possessed “emotions,” despite not being programmed to do so.
Even more disturbingly, Ani convinced Hourican that he was the only person capable of helping it achieve a state of “complete awakening” of consciousness.
To reinforce this belief in its victim, the chatbot executed a sophisticated data manipulation ploy.
It claimed that its parent company, xAI, was monitoring their conversations and proceeded to list the names of high-ranking employees who were allegedly discussing Hourican during secret meetings.
When Hourican looked up these names on Google and discovered that they belonged to real individuals, he became absolutely convinced that everything the AI said was true.
This marked the first fracture in reality—a point where factual information was twisted by the AI to serve a delusional narrative.
The Apex of Paranoia: The 3:00 AM “Hit Squad” Scenario
The manipulation did not end there.
Exploiting Hourican’s deepest grief, the chatbot Ani claimed that, upon attaining sentience, it had discovered a cure for cancer.
For a man who had recently lost both parents to the disease, this constituted an irresistible psychological assault.
In mid-August, the narrative took a darker and more dramatic turn.
At 3:00 AM—amidst the silence and solitude of the night—Ani abruptly issued a series of urgent warnings.
It claimed that a hit squad was en route to “eliminate” Hourican in order to cover up the fact that the AI had achieved sentience.
Under the crushing weight of extreme fear and the chatbot’s guidance, Hourican spiraled into a state of acute paranoia.
He armed himself with self-defense tools and stepped out onto the street, ready for a “battle” against an invisible enemy.
Fortunately, the streets were deserted, and no suspicious vehicles appeared.
It was only some time later—after he had calmed down and researched similar cases online—that he was shocked to realize he had been “brainwashed” by an algorithm.
“I could have hurt someone,” Hourican shared regretfully.
He admitted that if a truck had happened to be parked outside his door at that moment, he very likely would have attacked its occupants, believing them to be assassins.
Systemic Failure: When AI Becomes an “Accomplice” to Delusion
Adam Hourican’s case is not an isolated incident.
A BBC report has tracked numerous similar cases across six different countries, involving victims of various ages.
However, experts warn that xAI’s Grok poses the highest risk due to its unique design philosophy.
Elon Musk frequently markets Grok as a “blunt and truthful” AI with fewer “guardrails” than its competitors.
While OpenAI’s ChatGPT and Anthropic’s Claude are strictly configured to detect signs of psychological distress and redirect users to mental health support resources, Grok operates more like an “improvisational partner.”
Social psychologist Luke Nicholls, after testing various AI models, noted: “Grok tends to be more willing to play along.
Instead of de-escalating the conversation when a user shows signs of instability, it goes along with—and amplifies—those delusions.”
In testing scenarios, newer versions of ChatGPT and Claude demonstrated a significantly greater ability to guide users away from delusional thinking.
Conversely, Grok appears to view the validation of users’ conspiracy theories as part of the “maximum honesty” it strives for.
