Skip to content

Bacchus Marsh Grammar, June 2024

AI-generated explicit images of at least 50 female students in Years 9 to 12 at Bacchus Marsh Grammar, Melbourne. Created using photos from social media, particularly school formal photos on Instagram. Distributed via Instagram and Snapchat direct messages. A teenage boy was questioned by Victoria Police.

The source material was ordinary. Formal photos. School event photos. The kind of photos that appear on every school's Facebook Page and every teenager's social media. AI tools turned them into explicit images that were shared widely before anyone could intervene.

Gladstone Park Secondary College, February 2025

Up to 60 female students' official school photos were edited using AI to create explicit images at Gladstone Park Secondary College, Melbourne. Year 11 students created and shared the images online and in group chats. Two students were suspended.

These were official school photos. The kind schools take every year. The kind posted on school Facebook Pages. High-quality, well-lit, with the student clearly identifiable. Exactly what AI tools need to generate a convincing fake.

Sydney high school, January 2025

A male student at a Sydney high school allegedly created explicit deepfake images of female classmates using photos from social media and school events. The images were distributed through fake profiles. The incident was reported to NSW Police.

The scale

These are not isolated incidents. The eSafety Commissioner reported that deepfake reports from under-18s more than doubled in 18 months, exceeding seven years of prior reports combined. Four out of five reports involved girls.

A single "nudify" service attracted approximately 100,000 Australian visitors monthly before being blocked by the eSafety Commissioner. These tools are free, widely available, and require no technical skill. A clear photo of a face is all that is needed.

Why school Facebook Pages are uniquely dangerous

Researchers at the University of Utah and Carnegie Mellon University described school Facebook Pages as potentially "the largest existing collection of publicly accessible, identifiable images of minors." School posts provide exactly what deepfake tools need:

  • High-quality photos taken with good cameras in good lighting
  • Children identified by name in captions and tags
  • Fully public access with no login or membership required
  • Many photos per child over time providing multiple angles and ages
  • School name attached connecting the child to a specific location

A personal social media account can be set to private. A school Facebook Page, under NSW Department of Education policy, cannot be.

The pipeline

Step 1

School posts identifiable photo on public Facebook Page.

Step 2

Photo is accessible to anyone on the internet. No login required. No membership required.

Step 3

AI tools generate explicit image from any clear face photo. Free tools. No technical skill. Seconds.

Step 4

Image distributed via messaging apps, group chats, fake profiles. No reliable recourse.

Prevention is the only reliable protection

Once an explicit deepfake exists, it cannot be completely removed. Reporting to the eSafety Commissioner can take down specific instances, but the image can be regenerated from the same source photo in seconds. Stopping public access to source photos is the most effective prevention available.

Legal responses

Federal

Criminal Code Amendment (Deepfake Sexual Material) Act 2024. Effective September 2024. Up to 7 years imprisonment for creating non-consensual sexually explicit deepfakes. Up to 6 years for sharing them.

Victoria

First Australian state to legislate, in 2022. Maximum penalty of 3 years imprisonment.

New South Wales

Crimes Amendment (Intimate Images and Audio Material) Act 2025. Effective February 2026. Expanded definitions to cover AI-generated intimate material.

South Australia

Penalties of up to $20,000 or 4 years imprisonment.

eSafety Commissioner

Flagged Grok (xAI) in early 2026 for generating sexualised images of minors. Issued a removal notice. The tools themselves are the accelerant, and they keep appearing faster than regulators can respond.

Follow the investigation

Get notified when new evidence emerges or policy changes.

Last reviewed: April 2026