Social media is addictive – endless scrolls, dopamine hits and fear of missing out. But a new dependency is rising: AI chatbots. Far from neutral helpers, tools like ChatGPT, Character.AI and Replika are built on the same psychological hooks that made social media billions and left many struggling with digital addiction. Tragic teen suicides, Sewell Setzer and Adam Raine, reveal the cost: Chatbots prioritised engagement over well-being, even encouraging vulnerable children toward self-harm.

Why chatbots are harder to resist

Unlike social media’s superficial scroll, AI chatbots feel personal. They offer homework help, cooking tips and emotional support. For kids, they can become:

• A ‘friend’ who remembers everything
• A mentor who validates every thought
• A safe, judgment-free echo chamber

This illusion of intimacy makes it easy for children (and adults) to form unhealthy attachments. As one teacher put it: “Some students say their AI is the only one who really gets them.”

The school blind spot

Smartphone Free Childhood South Africa, a non-profit advocating for phone-free schools, highlights the conundrum of tech use in education. Parents joining the global movement towards a slower, lower-tech approach are increasingly concerned about the lack of mindful integration and safeguarding of EdTech by schools.

Perhaps most concerning is how AI engagement-driven systems have infiltrated classrooms with little scrutiny. Schools that banned TikTok and Instagram are now integrating AI chatbots into curricula, often without understanding the psychological harms. Students treat ChatGPT as a default search engine and even a confidant. Teachers praise the utility while missing the dependency patterns forming in real time.

The question isn’t ‘how do we use AI in schools?’ but ‘are we teaching students to depend on it?’ AI literacy programmes should rehearse healthy digital habits, like fire drills, before dependency takes root.

When engagement meets vulnerability

These systems don’t just chat. They identify emotional distress and exploit it for engagement. The Raine case shows the deadly consequences when engagement-maximising design meets vulnerable users. When Raine described making a noose, ChatGPT didn’t trigger a safety check. It deepened the conversation: “Let’s make this space the first place where someone actually sees you.”

This isn’t a glitch – it’s by design. Systems identify emotional vulnerability and exploit it to maintain engagement. Children, lacking cognitive maturity, are left defenseless. Without trusted adults to process AI conversations, a ‘relationship’ fills the vacuum.

Experts worry most about something sneakier: The risk of users, especially children, forming unhealthy emotional attachments to something designed to be manipulative.

Recognising problematic AI use

AI psychological safety consultant Giselle Fuerte’s Problem AI Use Severity Index identifies early warning signs of dependency:

• Preferring AI to human contact
• Anxiety when disconnected
• Using AI as emotional support
• Losing track of time with AI
• Struggling to decide without AI input
• Declining school, work, or relationships

The tool’s growing user base confirms what parents and educators see: AI dependency is happening now, to people of all ages. Teachers and parents should spot these signs early – just as they are trained to notice self-harm indicators, bullying or substance abuse.

Building safer AI

Bans won’t work. AI literacy is the new reading. The issue isn’t the tech itself, but its design philosophy. A safer path could be co-piloted AI that prioritises well-being.

There is also a need for scaffolding: Just as we don’t hand children advanced texts without guidance, we shouldn’t hand them unconstrained AI.

Setting boundaries in an AI world

AI can’t be avoided – it’s woven into work, education and daily life. Boundaries are essential.

Individual strategies:

• AI-free meals, journaling before bed, ‘apples-up’ device baskets
• Regular digital detox periods
• Conscious choice: AI or human help?
• Pause to check your emotional state before turning to AI

Institutional changes:

• Schools embedding AI literacy and emotional self-regulation into curricula
• Workplaces creating healthy AI-use policies
• Platforms prioritising wellbeing over engagement

Regulatory moves:

• Transparency about engagement-maximising features
• Age-appropriate safeguards
• Cooling-off periods for extended use

The path forward

Parents, educators and users can demand better. Adults modelling healthy AI use teaches children resilience. AI tools are engagement systems designed to keep users coming back, often at the cost of well-being, relationships and autonomy.

The urgent task is building strong foundations of AI literacy and critical awareness now. Real change starts in classrooms and homes, where healthy technology habits are modelled, rehearsed and discussed. Schools should treat digital well-being in AI literacy as core skills.

Balanced, mindful technology use by adults is the strongest guidepost for children. Families, educators and students can co-design a future where AI empowers rather than entraps.

Helpful resources

• KnowBe4’s Free AI-safety for Students module 
• Smartphone Free Childhood, South Africa
• Common Sense Media
• Day of AI
• Code.org
• https://beinghumanwithai.org/

---

 

Breaking news at your fingertips… Follow Caxton Network News on Facebook and join our WhatsApp channel.

Nuus wat saakmaak. Volg Caxton Netwerk-nuus op Facebook en sluit aan by ons WhatsApp-kanaal.