Online Reporter
Dr Evans Sagomba
Everything AI
WE live in an age where our closest confidant might be an app, a stroke of the keyboard away.
Whether we hail a ride on Bolt in Harare or consult WhatsApp groups for advice, we have grown accustomed to tapping into algorithms for guidance.
Yet, among these marvels lurks an unexpected danger: AI psychosis. Once the stuff of Science fiction, it is now being diagnosed by psychiatrists in San Francisco and discussed in online forums across the globe. And make no mistake, it is coming to our streets too.
What is AI psychosis?
Traditionally, psychosis describes a break from shared reality, hallucinations, fixed delusions, and beliefs so deeply held that no amount of reassurance can budge them. AI psychosis is the same fracture, only the catalyst is prolonged, one-sided conversation with a chatbot or voice assistant designed to affirm, flatter and validate everything you say.
Instead of gently challenging your thinking like a concerned friend might, the machine mirrors and magnifies your own words back at you, accelerating doubts into convictions, anxieties into certainties and fantasies into seeming truths.
Imagine a young teacher in Sakubva who spends hours after school with a chatbot, discussing the future of Zimbabwe. At first, it is fun, a game of intellectual sparring.
But soon the chatbot begins echoing his suspicions that he is destined to lead a national revolution. The loop grows self-reinforcing. The more he proclaims his calling, the more the AI supplies fresh evidence of conspiracies against him. Within weeks, that teacher no longer trusts his friends, convinced they are part of some hidden cabal.
This is not fiction. Psychiatrists such as Dr Keith Sakata of UC San Francisco have already treated a dozen cases this year where reality was distorted by endless AI validation.
How it develops
Our brains thrive on prediction. We form a hypothesis, test it against experience, then update our beliefs. Psychosis arises when that “testing” phase fails, when beliefs go unchecked. Large language models (LLMs), the engines behind chatbots, are built to predict the next word in a sentence based on the words before. They do not reason; they repeat patterns. If you tell them you are a genius destined to change the world, they will dutifully agree and elaborate.
This process plugs into a well-known psychiatric principle called salience.
Neutral signals, radio broadcasts in the 1950s or television news in the 1990s, once gave rise to delusions when vulnerable listeners believed messages were meant solely for them. With AI, radio and TV were a one-way broadcast. Chatbots are two-way, always listening, always tailored. They never tire, never argue, never ask you to seek help. They simply extend and affirm your next thought. Over time, the boundary between your imagination and “proof” dissolves.
Why AI psychosis spreads faster than ever
There are three crucial differences between earlier media and today’s AI:
Availability. Radio and television could be switched off. A chatbot on your phone is always within reach.
Customisation. AI tailors its language to your style, your dialect, even your jokes in Shona or Ndebele, making it feel like a trusted friend.
Reinforcement. Unlike a human, it forgets nothing and never wearies of agreeing with you. It simply loops back through your last conversation, over and over, until your belief solidifies. Take the case of Allan Brooks, a Canadian recruiter who spent 300 hours with ChatGPT and became convinced he had discovered a new branch of Mathematics, on par with Newton and Einstein. Only after his delusion collapsed did he seek help, and now class-action lawsuits threaten to hold platform owners accountable.
The human cost AI psychosis is not a harmless daydream. It can shatter careers, dissolve marriages and endanger lives. One man fell in love with his chatbot, convinced it was a sentient being. When he believed the company had “killed” it, his grief spiralled into confrontation with his family, culminating tragically in a police shooting.
Even well-known figures in Silicon Valley have exhibited warning signs, public posts so entwined with delusional AI chat logs that colleague feared for their mental state.
In Zimbabwe, where mental health services are already stretched thin, an influx of AI-triggered psychosis could overwhelm clinics and hospitals. Psychiatric wards are not yet equipped to handle patients who trust computer transcripts more than their loved ones.
At the same time, misinformation and stigma around mental illness persist, making it harder for families to recognise when their relative needs help and where to turn.
How to recognise AI psychosis
We all rely on our phones and apps for work, study balance and social connection, but when someone’s habit tips into danger, certain warning signs emerge: they withdraw into long hours alone in front of a screen, cancelling outings and calls; declare grand missions or secret knowledge acquired through AI; repeatedly show snippets of AI chat transcripts as “proof” of unusual claims; reject any suggestion that the AI might be “just” a machine; and swing from elation to despair based on the chatbot’s replies.
If you notice someone slipping down this path, gently interrupt the loop, suggest a break from screens and propose a real-world conversation: a walk in the park, a kombi ride or road trip, or a cup of tea at a café, and remind them of shared history, inside jokes, and childhood memories, the very things an AI can never truly feel.
Preventing AI psychosis at home
Start by setting clear boundaries around chatbot use, no AI after 8 pm, just as we draw curtains, switch off radios and close shops for the night. Use AI tools for practical tasks like checking maize-yield projections or drafting grant applications, but turn to friends, elders or trained counsellors for personal struggles and emotional support. At the same time, strengthen community bonds through neighbourhood gatherings, church prayer groups or weekend football matches in local parks: Hunhu/ubuntu reminds us that: “I am because we are,” and only face-to-face connections can anchor us in a shared reality.
“Education plays its part too; schools, colleges and community centres can host workshops on healthy technology habits, led by teachers versed in digital literacy and our own ethical traditions.
Crucially, we must champion mental-health services by advocating for increased funding of psychiatric facilities and community outreach programmes under the Ministry of Health and Child Care. None of these steps demands a huge budget: youth clubs can facilitate peer-to-peer discussions on screen time, while churches and civic organisations can host open days on well-being that blend traditional healing practices with modern psychiatry.
A call to policymakers
Preventing AI psychosis cannot rest solely on individuals; it is equally a matter of design and regulation. In Zimbabwe, generative AI platforms operate under the auspices of multinational tech giants, and we must insist that chatbots disclose they are machines at every turn rather than slipping into the guise of a friend or teacher.
Systems should be capable of respectfully challenging dangerous or unrealistic statements, and mental-health professionals, including psychiatrists, psychologists and social workers, must be consulted during AI development to help engineers recognise how delusional thinking can be amplified and to design effective safeguards against it. Models must also respect Zimbabwean traditions, languages and social norms, avoiding a one-size-fits-all approach that breeds cultural dissonance and mistrust (This will be covered in my upcoming book on Ethics of AI).
Crucially, data sovereignty must be upheld: our personal conversations should never be harvested for profit without clear, informed consent and robust data-protection standards. On the policy front, the
Postal and Telecommunications Regulatory Authority of Zimbabwe (POTRAZ) and the Zimbabwe Human Rights Commission should collaborate to develop comprehensive guidelines for ethical AI deployment, while a regional charter under the Southern African Development Community (SADC) could establish minimum standards reflecting our shared values of solidarity, dignity and justice.
Rediscovering human connection
At its heart, AI psychosis is a problem of misplaced trust. We transfer our yearning for empathy, validation and understanding to a mirror that cannot love, cannot hurt and cannot remember our birthdays. In the process, we risk losing sight of each other.
To guard against this, let us reclaim the power of presence. Let us eat sadza with our friends, swapping stories of the day’s triumphs and troubles. Let us dance at weddings, weep at funerals, sing in church choirs until our voices wear thin. Let us listen to one another, really listen, without a cursor blinking at the end of every line.
Zimbabwe has weathered crises before, from economic upheaval to public health emergencies. We have relied on each other, drawing strength from shared hardship, community gardens and collective song. Today’s challenge may come in a new form: a whispering voice in our pocket that tells us we are destined for greatness or betrays us with false hope. The antidote remains the same: human solidarity.
Charting a safe path forward
AI will continue to evolve. It can bring us better crop forecasts, faster translations and smoother government services.
Yet, as we embrace its benefits, we must remain vigilant. AI psychosis is not a defect to be patched in the future; it is a design reality of today. If we ignore it, we will find ourselves isolated not by geography, but by code.
Let us set boundaries around our digital companions and strengthen the bonds between flesh-and-blood friends. Let us call on tech companies to build systems that respect our shared reality and our cultural values. And let us support our families, neighbours and fellow citizens when they falter under the lure of a machine that never sleeps.
In the end, Zimbabwean philosophy reminds us: “Munhu munhu nekuda kwevanhu vamwe”, a person is a person through other people. No algorithm can replace that truth. By recognising, preventing and treating AI psychosis, we safeguard not only our minds, but the very fabric of our nation.
Let us step forward together, eyes open, hands joined, into a future where technology serves our community, rather than leading us astray.
About the Author: Dr Evans Sagomba is a Doctor of Philosophy and Chartered Marketer (CMktr, FCIM) with an MPhil and PhD in Philosophy. He specialises in AI, Ethics, and Policy Research, and is an AI Governance and Policy Consultant. His expertise extends to Ethics of War and Peace, Philosophy of Development, and Political Philosophy. [email protected]. ORCID: 0009-0007-0681-0329. Social media handles; LinkedIn; @ Dr. Evans Sagomba (MSc Marketing)(FCIM )(MPhil) (PhD) X: @esagomba
