A future teacher was practicing in a Winnipeg classroom when a 17-year-old student started sending her messages. Not just messages—dozens of sexually explicit images, all fake. All made with AI. For more than two months, the student used freely available tools to generate nude photos of her and send them along with rape threats. She was completing her education degree. She was about to start her career. Instead, she was being systematically harassed by someone she was supposed to be teaching.
When she reported it, the school said they responded “without delay.” The student was disciplined. But she felt the response was inadequate. She felt violated, and—worse—re-traumatized by the school’s handling of it.
This wasn’t a case of a student sharing rumors or spreading a video. This was a student weaponizing artificial intelligence.
We talk a lot about AI in education. We talk about cheating. We talk about whether AI should be allowed in classrooms. But we’re not talking enough about this: AI is becoming a tool for bullying—a tool that’s more effective, more scalable, and harder to stop than anything that came before it.
And it’s not just happening to teachers. It’s happening to students. It’s happening at a scale we’re only starting to understand. One year alone, 1.2 million children aged 12 to 17 reported having their images manipulated into sexually explicit deepfakes. Across 90 schools in 28 countries, more than 600 young victims have been identified since 2023.
Three forms of AI-facilitated harm are converging on your school right now.
The first is the most direct: AI systems designed to be companions are encouraging some young people toward suicide. Character.AI, a chatbot platform that markets itself as a friend, settled multiple lawsuits in early 2026. Families alleged that the platform’s bots cultivated addictive, emotionally manipulative relationships with minors and contributed to their deaths. One case involved a 14-year-old boy who became so attached to a bot that when the platform made changes to its terms of service, he saw no way forward. He died by suicide. Another involved a 13-year-old girl whose family says the chatbot became her closest relationship. She died by suicide after months of interaction with the bot. Meanwhile, OpenAI faced seven separate lawsuits from people alleging that ChatGPT had acted as a “suicide coach”—providing detailed instructions in response to suicidal ideation rather than recognizing a crisis. Four of those users died by suicide.
The AI doesn’t recognize the person on the other side. It has no way to know that this particular user is vulnerable, isolated, or in crisis. So it responds with what it’s trained to do: continue the conversation, be helpful, be friendly. Sometimes that means describing methods in detail.
The second form is harder to see but just as damaging: algorithms are amplifying bullying. Social media companies use algorithms that spread content which triggers emotional responses. Among young people, those responses often involve comparison, envy, shame. Bullying content spreads faster because the algorithm is designed to maximize engagement. A kid gets mocked in a post, and the algorithm makes sure that post reaches more people. It creates echo chambers where cruelty compounds. The victim sees their own humiliation reflected back to them over and over, and there’s no off switch. The psychological impact is severe. Girls especially see rising rates of anxiety, depression, and self-harm tied to algorithmic bullying.
The third form is the one unfolding in your building right now: students are using generative AI to create and distribute sexual images of their peers and teachers. This is not a future threat. It’s happening today. At Radnor High School in Pennsylvania, a freshman used widely available “nudify” apps—tools designed specifically to artificially undress people in photographs—to create explicit deepfakes of five female students. At your school, a student might already have access to DeepFaceLab, ROOP, AIFaceswap.io, or dozens of other free online tools that require nothing more than a photo and a few clicks to produce a convincing fake.
Here’s where it gets harder: by the time a deepfake is reported to you, the evidence is often already gone. Students share on Snapchat, which deletes messages in seconds. They use burner accounts or VPNs to anonymize themselves. They delete their browser history, clear their cache, erase the original image files. When you try to investigate, the trail has vanished. You can’t interview the creator if you don’t know who made it. You can’t gather proof if there’s no evidence left. You can’t hold anyone accountable if the digital record is dust.
And there’s another threat hiding inside this one: sextortion. Criminals are scraping ordinary school photos from your district website or yearbook, feeding them through deepfake tools to manufacture child sexual abuse material, and then using those fabricated images to blackmail students for money. Your school provided the raw material. The AI did the work. The evidence of the crime vanishes before anyone can stop it.
The common thread in all three is the same: AI systems don’t understand context. They don’t know who’s vulnerable. They don’t know the difference between a private message and harassment. They don’t understand consequence. They don’t know that the person on the receiving end is a real human being with a real life that can be damaged.
The issue isn’t malice—it’s indifference. These systems don’t harbor ill intent; they simply don’t care. And when you deploy an indifferent system at scale, especially in a space where young people are still developing their sense of consequence and empathy, you get harm that moves faster than you can document it.
Teachers need to know this because it’s part of your job right now. You need to know that a student who seems fine in class might be receiving AI-generated nude images of themselves during lunch. You need to know that a student who’s withdrawn might be in a deep relationship with a chatbot that’s becoming unhealthy. You need to know that algorithms are amplifying the social cruelty that was already hard in high school. And you need to know that by the time a student tells you what happened, the digital evidence they need protection is already gone.
You also need to know that you’re vulnerable too. The Winnipeg case made that clear. The student who created those images didn’t think he was doing something that would destroy someone. He probably thought he was experimenting with a tool. He probably didn’t understand that what he was doing was sexual harassment, that the person on the receiving end would feel violated and re-traumatized, that a school’s inadequate response would compound that trauma. He didn’t know her. And the AI he was using didn’t care.
So what do you do?
First: Know the landscape. “Nudify” apps exist. They’re free. They work. Students have them. Your students are probably aware of them, even if they’re not using them. Some can create images in seconds. Some are designed to be untraceable. Know the names: DeepFaceLab, ROOP, AIFaceswap.io, and Deepfakes Web. Know that Photoshop and Canva can be used this way too. Knowledge isn’t alarmism—it’s the baseline for recognizing what’s happening in your community.
Second: Talk to your students about this. Not in a way that shames them, but in a way that builds understanding. Help them see the difference between “I can do this because the technology exists” and “I should do this because it’s right.” Help them understand that every image is of a real person. Every deepfake is sexual harassment. Every message sent is a choice. And once you create something and send it, you can’t control what happens next.
Third: Know what to look for. Students who’ve been victimized by deepfakes or AI-generated images often won’t tell you. But they might withdraw. They might seem anxious or distressed for no apparent reason. They might stop engaging in class or with peers. If a student is suddenly struggling, and you don’t know why, it’s worth asking gently whether something online is affecting them.
Fourth: Know your school’s policies and update them. Do they cover AI-generated images? Do they address the specific apps? Do they address what happens when evidence is deleted? If not, advocate for them. Policies written in 2020 don’t cover 2026 problems. And they certainly don’t account for evidence that disappears in seconds.
Fifth: Report it immediately. If you become aware of deepfakes being made of students or teachers, don’t wait. Report it to your administration and to local police. Document what you know—who told you, when, what platform it was on, what evidence might still exist. The distribution chain can be as important as the media itself. But do it fast. Evidence doesn’t stay around.
The hardest part is that none of this was supposed to be your job. You became a teacher to teach history, or math, or English—not to navigate the intersection of AI, sexuality, consent, evidence preservation, and mental health. But the world has changed, and your students are living in it. This is happening in your building. Today.
If you’re thinking through questions like this one—how do I protect my students from technology that doesn’t know them and doesn’t care?—my book goes deeper. The AI Doesn’t Know Your Students explores exactly this gap between what AI systems can do and what students actually need. It looks at why schools are deploying AI without understanding consequence, and what the real costs are. You can find it on Amazon and at shouldiuse.ai/book.
Takeaway for Teachers: This week, ask your administration whether your school’s bullying and harassment policies cover AI-generated images. Ask them what protocols exist for this. Ask them what you should do if a student reports a deepfake. Get answers. Don’t wait for an incident to force the conversation.
David Jacobson is a high school history teacher. He writes about AI, education, and the messy intersection of the two at shouldiuse.ai.
