On a sunny morning in May, I had the opportunity to lead a unique and forward-looking class with a group of bright and curious 3rd graders - for the second time. The topic: Deepfake technology and the importance of digital awareness in the age of artificial intelligence. What unfolded was a session full of engagement, critical reflection, and even a touch of imagination, as we not only discussed the risks of AI-generated media, but also created an entirely fictional online persona, from scratch.

Understanding Deepfake: Starting the Conversation

We began with a simple yet important question: “Do you always believe what you see online?”

That question sparked a wave of discussion, as the students shared their experiences with social media, YouTube, and even some of the odd or funny videos they’ve encountered.

To ground the lesson, I explained what artificial intelligence is in simple terms - like a very clever robot that learns from what it sees and hears. We then introduced the concept of deepfake technology, using age-appropriate examples. I explained how AI can take someone’s face or voice and make it look or sound like they are saying or doing something they never did.

The students were amazed (and a little shocked) to see how realistic some examples could be.

Bringing AI to Life: Creating “Mirela Popescu”

To help them better understand how deceptive deepfakes can be, we built a fictional character together:

Mirela Popescu, a 53-year-old Romanian luxury car executive living in Manhattan with her billionaire husband, designer glasses, a glamorous wardrobe, and a very convincing Instagram presence.

Using AI tools like ChatGPT, Microsoft Copilot, and Perplexity, we developed her story, designed her Instagram bio, and imagined her family, vacations, and lifestyle.

Then came the twist: we generated her photos and videos with AI as well, using deepfake-style image synthesis. These posts, which looked completely believable, were shared on a freshly created Instagram account. The children were stunned by how “real” Mirela looked. This opened the door to a deeper conversation: If we could create someone so convincingly in a few hours, what about the content we consume daily online?

Capturing the Classroom: Observation and Interaction

We even used image analysis tools (like ChatGPT’s vision mode) to analyze a photo of the classroom. The model described what it “saw”: attentive children, an orderly space, and a calm atmosphere. This demo showed students how AI can now “read” images, but also how it can misinterpret context or be manipulated.

Together, we reflected on the responsibility of using AI, how to identify fake media, and why it’s always smart to ask questions - even about things that seem obviously true.

Tools Used During the Session

• ChatGPT and Sora – for narrative creation, vision analysis, and dialogue generation
• Perplexity.ai – for fast, AI-powered search results
• Microsoft Copilot – to help refine language, generate captions, and organize thoughts

A Lesson in Digital Literacy

This session wasn’t just about AI or deepfake, it was about critical thinking. About helping young students realize that they can (and must) pause before they share, believe, or react to what they see online.

They left the room not just with a fun story about a fictional millionaire from Manhattan, but with a stronger sense of responsibility, and a new curiosity about the world behind the screen.

And that, I believe, is one of the most valuable lessons we can teach in the digital age.