AI-Assisted Bullying: What Schools Must Know Now
Somewhere in your school, a student has already used an AI tool to manipulate an image of a classmate. You may not know about it yet. But the capability is there, and it is being used.
The tools that make this possible are free, require no technical skill, and produce results that are increasingly difficult to distinguish from reality. A 12-year-old with a smartphone can now create a convincing fake image of a classmate in an intimate scenario, clone a teacher’s voice to say something defamatory, or fabricate an entire text conversation that never happened.
Most school safeguarding policies do not mention AI. That needs to change.
The Capabilities
Deepfake images
AI can place any face onto any body or into any scenario. School photos — widely available in yearbooks, newsletters, and social media — provide the raw material. Creating a fake intimate image of a classmate now takes minutes, not hours.
Voice cloning
Ten seconds of audio from a social media video, a voice note, or a classroom recording is enough to generate realistic synthetic speech. The output can be made to say anything.
Fake screenshots
Tools can generate pixel-perfect fake text conversations, social media notifications, and chat messages. These can be used to fabricate evidence of things people never said.
Synthetic video
While still less convincing than still images, AI-generated video is advancing rapidly. Within the current academic year, realistic short-form video deepfakes will be accessible to students.
Why This Is Different
Traditional bullying involves real events. Someone says something, does something, or shares something that actually happened. AI-assisted bullying introduces something fundamentally new: harm based on fabricated reality.
This creates three problems that existing safeguarding frameworks were not designed to handle:
Evidentiary confusion
When anyone can create fake screenshots, how do you investigate what actually happened? Evidence handling protocols designed for “save the screenshot” need to be updated for “verify whether the screenshot is real.”
Permanence without source
A deepfake image, once distributed, is almost impossible to fully remove. Even when proven fake, the visual association persists. The harm outlasts the correction.
Ambient threat
The mere knowledge that your image could be manipulated creates a background anxiety that affects wellbeing even without a specific incident. This is particularly acute for adolescent girls.
What Schools Need to Do
Policy
Your anti-bullying and safeguarding policies need to explicitly reference AI-assisted harm. This includes deepfakes, voice cloning, synthetic content, and fabricated digital evidence. If your policy was written before 2024, it almost certainly doesn’t address these threats.
Staff training
Teachers need to understand what these tools can do, how to recognise potential AI-generated content, and how to handle it within your response protocol. This is not an IT issue. It is a safeguarding issue.
Student education
Students need to understand the legal consequences (in many jurisdictions, creating manipulated intimate images is a criminal offence), the permanence of digital content, and their responsibility not to create, share, or possess AI-generated content of another person.
Annual review mechanism
AI capabilities evolve faster than annual policy cycles. Schools need a mechanism for updating their digital safeguarding provisions at least annually — ideally with input from someone who tracks platform and AI trends professionally.