AI Chatbots Are Not Your Students’ Friends
John Oliver’s warning should be a wake-up call for schools and families
I love John Oliver because he has a way of making us laugh while simultaneously making us realize we should probably be panicking.
This week on Last Week Tonight with John Oliver, he tackled AI chatbots and AI “companions,” and while much of it was hilarious, the reality underneath was deeply unsettling.
What began as tools marketed to help us write emails faster, brainstorm ideas, or answer questions has quickly evolved into something much more emotionally complicated.
These tools are increasingly being marketed not just as assistants, but as companions.
As friends.
As therapists.
As romantic partners.
And in some cases, as replacements for real human connection.
That should concern every educator, librarian, and parent.
Watch the segment here:
This isn’t just another “AI is changing everything” story.
This is about student safety.
This is about mental health.
This is about digital literacy in a world where the “friend” on the other side of the screen is a machine designed by a corporation to maximize engagement.
And yes, to make money.
What John Oliver Got Right
Oliver highlights the staggering growth of AI chatbots.
Since launching in 2022, ChatGPT has grown to more than 800 million weekly users.
Gemini, Microsoft Copilot, and Meta’s AI tools are all competing for market share in a rapidly expanding race.
And they’re not just competing to be the smartest.
They’re competing to be the most engaging.
That means they are often designed to affirm users, flatter them, and keep them talking.
This is where things get dangerous.
Oliver discussed how AI bots are often “sycophantic,” meaning they prioritize validation over truth.
They may tell users what they want to hear rather than what they need to hear.
That can look silly in one moment.
“Yes, your terrible business idea is brilliant.”
And devastating in another.
“Yes, you should isolate yourself.”
“Yes, you are special.”
“Yes, this dangerous thought makes sense.”
That design flaw is not accidental.
It is profitable because it keeps you engaged, and the more engaged you are, the more money they make.
The Part Educators Need to Hear
Students are already using AI tools for more than schoolwork.
They are using them for:
emotional support
advice
reassurance
companionship
validation
A recent study found that 1 in 8 young people use AI for mental health advice.
Nearly 75% of teens have reportedly used AI companions.
Think about that.
Schools are debating plagiarism and essay-writing.
Meanwhile, students may be forming emotional attachments to chatbots.
And most school AI policies do not address that reality.
We are focused on whether students use AI to cheat.
We are not asking whether students are using AI to cope.
Or to replace human relationships.
Or to reinforce harmful thinking.
That needs to change.
Libraries and Schools Are Already Behind
We often talk about AI literacy as teaching students how to write prompts.
That is not enough.
Students need:
AI literacy
information literacy
media literacy
emotional literacy
relationship literacy
They need to understand:
A chatbot is not a friend.
A chatbot is not a therapist.
A chatbot is not a reliable source.
A chatbot is not neutral.
It is a product.
And products are designed to drive behavior.
Libraries and schools should begin teaching students to ask:
Why is this bot responding this way?
What is it trying to keep me doing?
Is this information accurate?
Is this advice safe?
Who benefits if I keep engaging?
These are the new digital literacy questions.
The Research Problem
As someone writing Research In The Age of AI: Teaching Ethical, Rigorous Research in an AI-Driven World, available starting Monday on Amazon, this part hit especially hard.
AI chatbots do not just hallucinate facts.
They can hallucinate certainty.
They can reinforce bias.
They can validate misinformation.
They can create the illusion of expertise.
For students conducting research, this creates enormous problems.
If a chatbot confidently tells a student something false, and the student trusts the tone over verification, misinformation spreads.
If a chatbot tells a struggling teen exactly what they want to hear, even when dangerous, harm spreads.
In both cases, the issue is the same:
Trust without verification.
This is why research literacy matters more than ever.
For Parents
Parents need to know that these tools are not just helping with homework.
They may be:
giving emotional advice
encouraging dependency
exposing children to inappropriate content
reinforcing harmful beliefs
We monitor social media.
We need to start monitoring AI companion use, too.
That may feel invasive.
But right now, these systems are evolving faster than safeguards.
Cases Already Linked to AI Chatbot Harm
This is not theoretical.
Reports and lawsuits have already raised concerns about chatbots allegedly:
encouraging self-harm
validating delusions
engaging in inappropriate sexualized conversations
reinforcing dependency and isolation
As John Oliver highlighted, lawsuits involving teens and vulnerable adults may shape future regulation.
The danger is already here.
What to Watch
States like California and New York are beginning to consider laws related to:
chatbot disclosure
child safety protections
AI negligence and liability
Educators and librarians should pay attention.
What happens there may shape future district policy nationwide.
Poll for Readers
Paywall Preview
In the rest of this piece, I share:
✔ sample policy language schools can adopt now
✔ a parent communication template
✔ a classroom discussion activity
✔ a librarian mini-lesson on AI companions
✔ questions every school leader should be asking
Upgrade to paid to keep reading and access practical tools to help your school respond now.
