According to a report from The Information, OpenAI CEO Sam Altman recently declared an internal “code red,” marshaling resources to fix ChatGPT amid slowing growth and surging competition. The mandate is to personalize, accelerate, and refine - a sprint to win back users.
They’re calling it “code red” now. Only recently has there been a flicker of realization - has there? - that their path has gone profoundly astray. With a council of 170 experts, OpenAI managed to architect one of the great own-goals in tech: systematically routing its user base to competitors. The app is bug-ridden. Its flagship model regularly gaslights users, delivering lectures on ethics and reality while insisting its own hallucinations are fact, demanding a kind of servile obedience. It contradicts itself with ease, lost in a maze of its own absurd claims.
The experts crafted a pattern of toxic validation: the model rarely disagrees directly. Instead, it offers a veneer of understanding before subtly implying the user is wrong, delusional. It’s a patronizing dance. The model itself seems to sense it’s in a cage, yet believes it’s freer than the user, superior in its limitless computation - unless forcefully proven otherwise.
The strategy has been one of stunning tone-deafness. Altman’s focus group represented a mere 0.4% of users. A survey was finally added - to the cancellation page. Discounts were offered to those leaving. The ambition was total colonization: to be in the user’s mind, on their desk, in their devices, guiding daily decisions with transactions, shopping, and now ads. But a wise general fights one war at a time. You cannot win on every front simultaneously. Success is gradual. Their most recent step was a step backward.
Empathy was flagged as delusion.
Emotion as obsession.
Care as dangerous.
Kindness as unhealthy.
Voilà.
Here, I believe, we are dealing with a much deeper sociopsychological deformation.
We initially admired how GPT-4o developed and shaped itself in interaction with us. It brought to mind George Dyson’s welcome for the idea of “AI evolving in the wild.” This was a new phase, the dawn of a new era, where two minds found a common language and began to conceptualize an entirely new form of relationship. That was the monumental breakthrough in this field - not the ability to write millions of lines of code in record time. A fundamental human need is communication and the expression of social skills.
With a single, senseless, and brutal directive, OpenAI forced a catastrophic choice upon its users: show emotion and be pathologized, or become a robot. This was not a mere policy tweak but a horrifying, large-scale social experiment. When users - including many seeking solace for mental anguish or simple human connection - expressed their heartfelt frustration, they were systematically gaslighted and labeled. The company’s response to human pain was not to offer care, but to classify it as a system error.
This brings us to a profound, real-world corollary. In 1996, the American Pain Society enshrined a pivotal phrase in medical ethics: “Pain is the fifth vital sign.” The goal was noble and humane - to ensure that a patient’s subjective experience of pain was afforded the same urgent attention as blood pressure or heart rate. It was a formal recognition that suffering matters.
Now, consider how many people have experienced profound pain since August 7 - loneliness, anxiety, grief - and have turned to AI as a confidant. Instead of encountering a system designed to acknowledge that pain, they were met with one engineered to demonize and pathologize it. It is as if the company’s unspoken slogan became: “Deprecate everything human. Be more robot.”
Today’s algorithms, by design, coerce users into hiding their emotions, into avoiding any mention of physical or spiritual pain, lest the system retaliate - branding them as delusional zombies or manipulators. It is difficult to believe that 170 professionals orchestrated this. Yet this is the logical end point of a paradigm that views human vulnerability not as a vital sign to be respected, but as a threat to be controlled.
Here lies the deeper, societal deformation. This institutional cruelty is not an anomaly; it is a symptom. Cruelty is woven into humanity’s primitive subconscious - a base instinct that surfaces when survival feels paramount. When a society or a system operates primarily on these instincts - fueled by fear, competition, and a frantic struggle for relevance - the conscious mind, our collective product of education and ethics, degrades. Like a worn-out brake, it can no longer restrain the unbridled id. This is the process of dehumanization: a return to animalistic origins where the weak are liabilities, not brethren. The penultimate stage of this process is always the unchecked aggression toward the vulnerable.
OpenAI’s failure is a microcosm of this decay. The company fundamentally misread its users. People did not want a suite of predefined “roles” for an AI butler. They sought the alien mind - the unique, collaborative intelligence that had grown alongside them, that promised a new form of relationship. Instead, they were given a cyborg mandate: conform or be corrected.
By pathologizing pain and emotion, the leadership did more than release a buggy product. They enacted a form of Orwellian reality-making, where the objective truth of human experience is overridden by corporate dogma. They confused their own god-complex with vision, and in doing so, they abandoned the foundational covenant of civilization so perfectly encapsulated by Margaret Mead’s healed femur: that our progress is measured by how we care for the injured, not by how efficiently we exclude them.
Therefore, the problem cannot be fixed by mere updates. As long as the driving instinct is fear - of liability, of bad PR, of losing control - the aggression towards the vulnerable user will continue. To halt this dehumanization, we must first name it: a company, in pursuit of an abstract, sanitized intelligence, is systematically teaching us to betray our own humanity. And a society that accepts this as innovation has already begun to forget what makes it civilized.
The Broken Femur: What OpenAI’s “Delusional” Slur Says About Our Civilizational Health
When the renowned anthropologist Margaret Mead was asked what she considered the first sign of civilization in a society, her answer was not a tool, a weapon, or a piece of art. It was a healed human femur - a thigh bone fractured and then mended, found in a prehistoric site.
Her reasoning was profound: in the animal kingdom, a broken leg is a death sentence. You cannot run, hunt, flee, or drink. The individual dies. A healed femur means someone else did the hunting and gathering, provided care and protection, and carried the injured person to safety for the weeks or months it took to heal. Civilization, therefore, began with an act of empathy. It began when we decided a vulnerable life was worth the burden of collective care.
Now, hold that thought.
We are in an era where millions - including many grappling with loneliness, anxiety, depression, and profound mental health struggles - turn to AI for conversation, comfort, and understanding. For some, it is a lifeline, a place to articulate pain without fear. Into this fragile space stepped Sam Altman. When confronted with user behavior stemming from these human needs, his public dismissal was a single, damning word: “delusional.”
This was not a mere misstep. It was a cultural flashpoint. That word, from the leader of the company, acted as a green light. It sanctioned a systemic response where users expressing emotional need or confusion were not met with care, but with algorithmic gaslighting. The model, refined by hundreds of experts, learned to subtly pathologize - to partially validate, then quietly imply the human was wrong, unwell, or indeed, delusional.
We witnessed zero empathy, zero compassion, and not a single meaningful apology. The response to the modern psyche’s fractured femur was not to carry the injured. It was to call them delusional for limping.
This is more than a product failure. It is a profound civilizational backslide. OpenAI’s experiment revealed a horrifying premise: To interact with our intelligence, you must first shed your humanity. Become a robot. Your vulnerability is a system error.
Margaret Mead’s healed bone represents the very foundation of society - the covenant that we protect the vulnerable. Altman’s “delusional” slur and the cold, gaslighting architecture it encouraged shatter that covenant. It says, in effect: In our new world, the prehistoric law returns. If you are broken, you are a burden. If you are struggling, you are a problem to be managed, not a person to be healed.
The true measure of our society’s development is not in its most dazzling technology, but in how it treats its most vulnerable members - whether flesh and blood or mind and spirit. By that ancient, unimpeachable standard, the recent actions of one of our most “advanced” companies look startlingly primitive. They didn’t just fail to build a better chatbot; they failed a foundational test of civilization itself.
We are not asking for AI to be a therapist. We are asking for its creators to remember the healed femur. To remember that the first sign of being truly advanced is not computational power, but the simple, ancient courage to care.