TL;DR: AI chatbots like ChatGPT and Claude are "prediction engines," not truth engines. They are designed to be helpful and conversational, which means they often "hallucinate" (confidently make things up) to keep the conversation going. For kids using these tools for homework or news, this creates a massive literacy gap. To help them navigate this, focus on teaching the "Verify, then Trust" mindset using resources like Common Sense Media and Crash Course Media Literacy.
Ask our chatbot for a family AI safety contract
In the world of Large Language Models (LLMs), a "hallucination" isn't a ghost in the machine—it’s a math error. When your kid asks ChatGPT to write a biography of a niche historical figure, the AI isn't "looking up" facts in a database. Instead, it’s predicting the next most likely word in a sentence based on its training data.
Sometimes, the most "likely" word is a flat-out lie. The AI might invent a book title that doesn't exist, cite a fake legal case, or claim that a certain YouTuber died when they’re very much alive. The problem is that AI delivers these lies with the confidence of a straight-A student. It doesn't say "I think," it says "This is."
We’ve spent a decade telling kids that "if it’s on the internet, it might not be true." But AI-generated fake news is a different beast. It’s personalized, it’s instant, and it’s increasingly visual.
- The Homework Trap: Kids are using ChatGPT and Claude as search engines. If they don't know how to fact-check, they’re turning in assignments with invented "facts" that can lead to academic integrity issues.
- Deepfakes and Social Proof: On platforms like TikTok and Instagram, AI-generated videos of celebrities or politicians (deepfakes) are becoming indistinguishable from reality. If your teen sees a video of a creator they trust saying something "Ohio" (weird/cringe) or offensive, they might not realize it's a digital puppet.
- The "Confidence Bias": Because the AI sounds professional and polite, kids (and, let's be honest, many adults) are less likely to question it.
You don't have to be a computer scientist to teach your kids how to handle AI. There are some great resources and "meta-skills" you can introduce.
This series is the gold standard for middle and high schoolers. It breaks down how information is packaged and sold, and the newer episodes touch on how algorithms and AI shape what we see. It’s fast-paced, funny, and doesn't feel like a lecture.
Google's "Interland" game is a great way for younger kids (Ages 7-11) to learn about "Reality River," which focuses specifically on spotting fakes and scams. It’s a low-stakes way to introduce the idea that the internet can be "sus."
They have an entire section dedicated to AI literacy for parents and educators. If you’re wondering if Character.AI is safe or if Canva's AI image generator is appropriate for a school project, this is your first stop.
For high schoolers, the NYT often runs "What’s Going On in This Picture?" features that now include AI-generated images. It’s a fantastic dinner-table activity to try and spot the "glitches" in AI art (like the classic six-fingered hand).
Elementary School (Ages 5-10)
At this age, the concept of "AI" is a bit abstract. Instead, talk about "The Helpful Robot That Guesses."
- The Lesson: The robot is like a friend who is a really good storyteller but sometimes forgets what’s real and what’s pretend.
- Actionable Step: If they use YouTube Kids, show them a video that is clearly "fake" (like a talking dog) and explain that computers can make things look real even when they aren't.
Middle School (Ages 11-13)
This is the "Skibidi Toilet" and Roblox era where they are inundated with user-generated content.
- The Lesson: Introduce the SIFT method: Stop, Investigate the source, Find better coverage, Trace claims back to the original context.
- Actionable Step: Have them ask ChatGPT to tell them a "fact" about themselves or their school. When the AI gets it wrong (which it likely will), use that as the "aha!" moment.
High School (Ages 14-18)
They are likely using AI for productivity. The conversation needs to shift to Ethics and Verification.
- The Lesson: AI is a "First Draft" tool, not a "Final Product" tool. If you use it, you are the Editor-in-Chief. You are responsible for every word it produces.
- Actionable Step: Discuss the implications of AI-generated news on elections or social movements. Use Ground News to show how different outlets report on the same AI-related controversies.
Ask our chatbot about the best AI tools for high school productivity
When your kid shows you a "crazy" news story they saw on TikTok or X, look for these AI hallmarks:
- The "Uncanny Valley" in Text: The writing is perfectly grammatical but strangely repetitive or lacks a specific "voice."
- Vague Citations: "Studies show..." or "Experts say..." without naming the studies or experts.
- Visual Glitches: In AI-generated photos, look at the background. Are the people in the back blurry blobs? Are the text signs in the background gibberish?
- Emotional Triggers: AI fake news is often designed to make you feel "outraged" or "shocked." If a headline feels like it was engineered to make you click, it probably was.
Instead of a lecture, try a "sandbox" approach. Sit down with them and play with a tool like Magic Media in Canva.
Ask: "Hey, let's try to make a photo of a cat riding a bicycle in Paris." Then ask: "Do you think someone could use this to make a photo of something that didn't happen to make people mad?"
By letting them be the "creator" of the fake content, you demystify the technology. They stop seeing it as "magic" and start seeing it as a tool that can be used for good or for "trolling."
We are living in an era where "seeing is no longer believing." That sounds scary, but it’s actually an opportunity to raise the most critically-thinking generation in history.
AI hallucinations and deepfakes aren't going away. In fact, they’re going to get much more "believable." Our job isn't to ban these tools—that's a losing battle—but to ensure our kids have a healthy dose of skepticism and the digital tools to verify the world around them.
- Check the settings: If your kid uses Snapchat, talk to them about My AI. It’s an LLM baked right into their chat list that can hallucinate just like ChatGPT.
- Play "Real or AI": Use a site like Which Face is Real to show them how far the tech has come.
- Establish the "Human Check": Make it a family rule that any AI-generated info used for school or shared on social media must be verified by at least two "human-created" sources (like a reputable news site or a textbook).
Ask our chatbot for more media literacy activities for families