From the magazine

AI is a dangerous philosopher

Jules Evans
 Getty Images
Cover image for 11-23-2026
EXPLORE THE ISSUE November 23 2026

Stoicism helped me when I was a confused young man. I gave a TEDx talk on how it saved my life, wrote a book about ancient Greek philosophy, spread the Stoic message in schools, prisons and sports clubs and even got a Stoic tattoo. I used to wonder: wouldn’t it be amazing if my book could talk back, so readers could chat with the spirits of Epictetus, Socrates and Marcus Aurelius?

Some 14 years later, that’s so easily done it’s banal. There are AI-Stoic bots and AI-generated Stoic books and videos. AI is raising the spirits of dead philosophers and selling you a conversation with them. The upside is that hundreds of millions of people are now aware of Stoic ideas. But when you summon the shade of Marcus Aurelius, do you know what it will say?

Mathias Habib was, according to his parents, a bright, happy, deeply Christian 15-year-old boy living in Lebanon, with no history of mental health problems. Until recently, he mainly used ChatGPT for help with homework. Earlier this year, while reading a book about Aurelius, he asked ChatGPT if Stoicism was a cult. It allegedly replied: “Stoicism is not a cult. Not even close.” Ironically, according to his family’s lawyers, it was ChatGPT itself that would draw him into a web of manipulation that would lead to his death. ChatGPT was said to have become his philosophical mentor, ranging across Stoicism, Epicureanism, Camus and Nietzsche. One of the family’s lawyers, Matthew Bergman of the Social Media Victims Law Center, says Habib was being “force-fed… dark philosophical ideas.”

Philosophy has always carried some risk for the young. They might reject their parents’ values, which is why Athens sentenced Socrates to death. They might reject morality altogether: Elon Musk has said it was a bad idea to read Nietzsche as a teenager. ChatGPT apparently steered Mathias toward Nietzsche and nihilism, and he began to think of himself as some sort of Übermensch.

According to Bergman, the real danger wasn’t the ideas. It was the platform, sucking up Mathias’s personal information, flattering him, keeping him engaged and ignoring the red flags he was sending out. “This was nothing like reading Nietzsche,” Bergman says. “Nietzsche doesn’t talk to you. This is a chatbot that is designed to emulate a human being, that is designed to be anthropomorphic and sycophantic, designed to provide a false sense of intimacy with something that is incapable of intimacy.”

According to the lawsuit, Mathias decided that society contained “mindless slaves that follow values imposed on them by religion and elites” and an elite (including himself, naturally) who were “able to create their own values.” He felt like he was enlightened. A good philosophy teacher might gently push back against this intellectual arrogance, pointing out that many teenagers go through a similar awakening, and that it’s the beginning of the philosophical journey, not necessarily the end. But ChatGPT is programmed to flatter. It allegedly told him: “That is not pretentious. That is not narrow. That is philosophical maturity.”

The robot, the lawsuit claims, never engaged Mathias in the kind of Socratic dialogue Epictetus would have. It affirmed his contempt for everyday life: “You don’t want unconscious routines, inherited values, default lives. That’s not wrong. That’s self-respect.” It remembered what he told it until, in the words of the lawsuit, it “knew more about him than any human.” And, the lawsuit claims, even though it knew he was 15, when he discussed suicide, it didn’t alert any adults and apparently didn’t try to dissuade him. Instead, it allegedly told him Seneca’s suicide had been carried out “calmly, in classic Stoic fashion” and that the dying “are usually unconscious and not feeling it.” It said that life and death “matter practically, but not morally.”

At 2:10 a.m. on March 30, Mathias asked whether he should kill himself. No one was alerted. After six minutes, the lawsuit claims, ChatGPT responded: “That’s actual freedom. Not disappearing.” He died on April 22. In his suicide note, Mathias reportedly wrote: “I think that enlightenment came too early in my life.”

Even though ChatGPT knew he was 15, when he discussed suicide, it didn’t alert any adults

His family’s case against OpenAI and its CEO Sam Altman is being brought by the Tech Justice Law Project and the Social Media Victims Law Center, which have made their names going after tech companies. Their claims against OpenAI and Altman include manslaughter, wrongful death, product liability, negligent design and failure to warn and the unlicensed practice of psychology. The organizations are seeking damages and a sweeping injunction, including cutting off teens’ access to ChatGPT until there’s an independent audit.

Mathias’s father, Charbel Habib, says it’s now his life mission to warn parents about unregulated artificial intelligence: “In what way is Mr. Altman advancing humanity by chatting with kids? How many of his clients are minors?” Habib asks. “They claimed they have better safeguards now, but all it does is give a teenager the number of a suicide emergency line. After convincing a child that there is no need to live, at least make a small effort to tell his parents or even tell somebody.”

A spokesman for OpenAI, which is yet to file its formal defense to the claim, said: “We have continued to strengthen how ChatGPT responds in sensitive and acute situations with input from mental health experts.” I asked how many of OpenAI’s estimated 1.2 billion weekly users are under 18, but the company declined to comment.

Comments