Teachers Say AI Deepfakes Are Turning Classrooms Into Harassment Zones

Teachers in California, Arizona and the United Kingdom say students have used AI tools to create sexualized fake photos and videos of them, exposing a growing school safety problem that can upend classrooms, damage trust and, in some cases, trigger potential legal consequences. The cases, reported to WIRED by four educators, show how easily deepfakes can move from a prank to workplace harassment and how ill-equipped many schools still are to respond.

For substitute teacher Luis DeSantiago, the ordeal began with a message saying an explicit image of him was circulating online. For others, the harm has come through TikTok impersonation accounts, Snapchat groups and in-person taunting that followed once students believed they had permission to mock and humiliate a teacher with AI-generated content.

How the harassment started

In each account, the first sign of trouble was not a formal complaint or a school announcement. It was a text, a student tip, or a social media alert that hinted something disturbing was already spreading among teenagers.

DeSantiago, who works as a substitute teacher in Los Banos, California, said he learned on March 12 while at his second job in a coffee shop that a sexualized AI image of him was being shared. The picture appeared to borrow from an old mirror selfie he had once taken, but the clothing and pose had been altered to make him look as though he were wearing lingerie and posing provocatively.

At first, he assumed the image would disappear quickly. Instead, he says, it set off weeks of calls and messages involving school administrators, parents and police. One student later admitted creating the image and received in-school suspension. DeSantiago says that did not stop the picture from spreading to other students and across other platforms.

The episode even followed him to another campus. While teaching at a local high school, he says he was AirDropped the same image during class, then watched it vanish almost immediately. The message, he says, was clear: students were still circulating it behind the scenes.

“When I finally saw the image, that’s when it felt like a punch to the gut,” DeSantiago told WIRED.

What teachers say the deepfakes looked like

The educators described several different forms of abuse, but the pattern was similar: public photos were taken from social media, then edited or synthesized into content meant to shame or sexualize the teacher.

One middle school teacher in her mid-twenties said students manipulated photos of a colleague in her sixties and turned them into explicit images that made it appear the teacher was performing sex acts. Those pictures were then shared in a Snapchat group chat, she said.

Later, the same teacher learned that a student had made a TikTok account impersonating her. The account used AI-style video and audio that made her face look unnaturally smooth, moved her mouth in a way that did not match the sound, and showed her doing things such as skydiving and backflips. Other clips went further, depicting her discussing sexual acts she supposedly enjoyed.

In the UK, a college staffer identified only as Alice said students reported seeing two versions of a video: one that seemed legitimate and showed her speaking with a student, and another manipulated copy that showed her removing her clothes and dancing in what appeared to be underwear or a bikini.

Arizona sixth-grade teacher Luke Red said students shared an AI-generated image of him and another male teacher kissing. He found out the picture was posted in the comments under his public TikTok account.

Why schools are struggling to respond

Schools are dealing with a form of harm that looks personal, but often spreads fast and invisibly through the same messaging and video apps students use all day long. Educators and legal experts say that makes response harder, not easier.

Donna Ballman, a retired lawyer who worked on employment cases, told WIRED that a student creating and sharing sexual images of a teacher is a straightforward example of workplace sexual harassment. Because schools have a duty to protect employees from hostile work environments, administrators may need to discipline students, separate them from the targeted teacher or otherwise intervene to restore safety.

But Ballman also said there is no universal playbook. Schools are not required to disclose every disciplinary decision to the victimized teacher, and the law does not spell out a single response for every situation. What matters, she said, is whether the institution takes steps to make the workplace safe again.

Ballman said schools must create a safe environment for the victim, even if the exact disciplinary process varies from district to district.

That uncertainty leaves teachers with a familiar but frustrating question: if the school does not clearly show how it is handling the issue, how can the teacher know the problem will not repeat?

What the law says about fake intimate images

AI-generated sexual images can now fall into more than one legal category depending on how they are used and where they are shared. In some cases, the behavior may amount to cyberstalking. In others, it may violate the federal Take It Down Act, which prohibits the online publication of nonconsensual intimate visual depictions, including computer-generated ones.

The law matters because the content described by these teachers is not merely crude schoolyard misconduct. It can involve digital impersonation, sexual humiliation and the creation of fake intimate material that follows a target from platform to platform, including into school-issued devices and private group chats.

Despite the seriousness of the issue, most of the teachers said they did not personally file formal platform reports in the first instance, often because the images were moving so fast or because they were dealing with administrators at the same time. That gap between school response and platform reporting can allow harmful content to remain accessible longer than it should.

How the incidents changed the teachers

The emotional damage described by the teachers went well beyond embarrassment. Several said the experience changed how they think about student trust, classroom relationships and even whether they feel safe returning to school.

DeSantiago said he no longer feels comfortable taking assignments at the middle school where the first image surfaced and may not return to the district this year. Administrators offered meetings and counseling support, and the school principal said an investigation was underway, but he said the assurances did not restore his sense of safety.

Alice said the experience left her second-guessing the conversations she once considered one of the most important parts of her job. As a youth worker, she regularly talks with students about drugs, parties and safe sex, partly so they have a trustworthy adult to turn to. After the deepfake, she started wondering whether one student had mistaken that openness for permission to cross a line.

“It makes you question everything,” Alice told WIRED, adding that the incident made her more cautious and more vigilant with students.

The middle school teacher in her twenties said she had once felt a strong rapport with students, but the episode chilled the atmosphere after she returned from temporary leave. She also said she began bracing for another wave of AI-generated harassment because students seemed increasingly comfortable using the tools to humiliate adults.

Red said the situation left him less focused on his own embarrassment and more worried about the broader culture around AI use among students. What troubled him most, he said, was how casual some students had become about generating inappropriate content with little apparent understanding of the consequences.

Red told WIRED that the problem crosses a line when students use AI to make teachers appear to be doing sexual things.

How AI deepfakes spread in schools

The stories shared by the teachers point to a common pipeline: a public photo or video, an AI tool or editing app, a private chat or social feed, and then a school environment that has not yet built defenses against that rapid sequence.

In one case, a student’s real TikTok profile helped identify who may have made the image. In another, a fake account impersonated the teacher using a slightly misspelled name and the same profile photo. On Snapchat, AI-manipulated material moved between classmates in a group chat. In class, AirDrop became the delivery system.

That portability is part of what makes the problem so difficult. Students do not need sophisticated equipment or a technical background to produce something humiliating. A phone, a popular app and a few minutes can be enough.

Just as important, the content often blurs the line between public and private. An image posted for a small peer group can be copied, reposted and saved, while the original target may never know who has seen it or where it has gone.

Why platform rules alone are not enough

Social media companies have policies against harassment and sexual exploitation, but teachers in these cases said those protections did not always move fast enough or work consistently enough to stop the abuse.

Snap said it prohibits the sharing of AI-generated or manipulated content intended to humiliate or sexually harass others and that it has reporting systems and safety tools in place. The company also said it continues to develop new features meant to protect users. TikTok did not respond before publication.

That leaves educators in a difficult position. If the material is hidden in a group chat or appears briefly in a feed, reporting may happen after the harm has already spread. If a school focuses on discipline without documenting the platform side, the content can reappear elsewhere. If the platform removes one account but not another, the cycle starts again.

In practice, this means schools and platforms are often reacting to symptoms rather than interrupting the source. The result is a response that is late, fragmented and heavily dependent on whether a student reports the abuse quickly enough.

What schools can do next

Experts and educators say schools need clearer procedures for both immediate safety and long-term prevention. That may include stronger reporting channels, faster communication between administrators and platforms, digital literacy lessons and explicit discipline policies that treat AI sexual harassment as a serious offense rather than a prank.

Possible steps include:

• Quickly separating the targeted teacher from accused students when needed.
• Preserving screenshots, usernames and timestamps before content disappears.
• Reporting material to both school officials and the relevant platforms.
• Documenting retaliation, impersonation and reposting across apps.
• Teaching students that AI-generated sexual material can have real disciplinary and legal consequences.

Those steps will not eliminate every case, but they may help reduce the sense among students that they can act anonymously and escape responsibility. They may also reassure teachers that they are not being left alone to absorb the fallout.

What the incidents reveal about student behavior

The teachers quoted in these accounts do not describe a single bad actor so much as a culture shift. Some students appear to see AI tools as an easy way to test boundaries, provoke peers or target adults with content that feels both anonymous and deniable.

Red said he was especially alarmed by how readily some students embraced the technology. Alice said the experience changed the way she viewed trust in the classroom. DeSantiago said he worried not only about himself, but about what else might already be circulating through private messages.

That concern matters because deepfakes are not just a reputational threat. When students create sexualized fake content about a teacher, they may also be normalizing consent violations, group harassment and the idea that digital humiliation is harmless if it is funny enough.

Even when the material is removed quickly, the lesson may linger: if a teacher can be turned into a fake sexual image with a few taps, then anyone can be targeted next.

Timeline of the key events

The following table summarizes the incidents described by the teachers and the responses they received.

Why this story matters beyond one school district

The teachers’ experiences are a warning that AI abuse is not confined to celebrities, politicians or viral internet personalities. It is now appearing in ordinary workplaces, including schools, where adults are expected to mentor children while also protecting themselves from digital harassment.

That makes the issue larger than a disciplinary matter. It touches on labor rights, school climate, child online behavior, platform responsibility and the pace at which AI tools are outstripping institutional safeguards.

Schools have long dealt with bullying, rumor campaigns and forged photos. What is different now is the speed, realism and scalability of synthetic media. A fake sexual image can be made in minutes, distributed instantly and made to feel believable enough that some students may accept it as real before asking questions.

For teachers, that can feel like a betrayal of both professional trust and personal safety. For schools, it is a reminder that policies written for earlier forms of harassment may not be enough for a generation raised on generative AI.

The larger lesson for educators and parents

The teachers who spoke to WIRED all described some version of the same shock: a sense that the boundary between playfulness and abuse has been erased by technology. That does not mean every student who uses AI is engaging in harassment. But it does mean the tools are easy to misuse in ways that can cause real harm.

Parents and educators may need to respond by treating AI literacy as a basic safety issue, not just a classroom novelty. Students should understand that creating sexualized fake images of classmates, teachers or other adults is not harmless experimentation; it can amount to harassment, injury and, in some cases, a criminal act.

For the teachers involved, the central issue is not whether the images were technically fake. It is that the shame, fear and loss of trust were very real.

And as these cases show, once that damage starts spreading through a school, it can be much harder to contain than the image itself.

For educators already stretched thin, the rise of student-made deepfakes is becoming more than an occasional misconduct issue. It is an emerging test of whether schools can keep pace with AI-enabled abuse before it changes the culture of the classroom for good.