The technology that lets anyone conjure a convincing image from a text prompt has arrived in one of the most sensitive corners of Australian life: the schoolyard. A round-up flagged by the OECD AI Policy Observatory points to a pattern that has become grimly familiar to principals, parents and police across the country, with AI-generated deepfakes being used to target school communities, in many cases by fabricating explicit images of students and staff.
It is a story that has been building for more than a year, and one FluentSea has tracked through its ugliest local expressions. Police have investigated explicit AI images linked to a Gold Coast teacher, an Adelaide man has been charged over AI-generated child abuse material, and a Perth oncologist was caught up in a case involving AI-generated child abuse videos. What the OECD framing adds is the wider point that these are not isolated incidents but a systemic problem now knocking on the doors of ordinary state and independent schools.
Why the schoolyard became a target
The mechanics are depressingly simple. Free or near-free image and video generators, many trained on vast troves of scraped internet photos, can take a single social media snapshot and produce fabricated nudes or sexualised video in minutes. The barrier that once protected people, the need for technical skill or expensive software, has collapsed. A teenager with a phone and a grievance can now do what would have taken a specialist studio a decade ago.
Schools are fertile ground for this because they combine three things: a dense network of young people who live much of their social lives online, a steady supply of publicly posted images, and the ordinary cruelties of adolescence. When a fake is created and shared through group chats, it spreads faster than any teacher can intervene, and the harm to the person depicted is immediate and lasting, regardless of whether anyone truly believes the image is real.
The law is catching up, unevenly
Australia is not starting from nothing. In 2024 the federal parliament passed laws creating serious criminal offences for sharing, and threatening to share, non-consensual sexually explicit material that has been digitally created or altered, closing a gap that had left many deepfake victims without clear recourse. The country’s online safety regime, overseen by the eSafety Commissioner, also gives victims a formal avenue to demand rapid removal of intimate images, including synthetic ones, and to escalate against platforms that drag their feet.
Yet the enforcement picture on the ground remains patchy. Where the alleged offender is a child, the response sits awkwardly between criminal justice, child protection and school discipline, and no jurisdiction has fully settled how to handle a 14-year-old who has generated abusive material of a classmate. Prosecutors must weigh the gravity of the harm against the reality that the perpetrator is also a minor. That tension is one reason cases have surfaced as a jumble of police investigations, expulsions and quiet settlements rather than a coherent national approach.
Two views on what to do next
Broadly, two camps have formed. One argues the pressure has to fall on the technology itself. On this view, the platforms and model developers that make image generation frictionless should carry more of the burden, through mandatory safety testing, watermarking of synthetic media, age assurance and faster takedown obligations. Supporters point out that expecting individual schools to police a problem manufactured by global tech companies is neither fair nor realistic, and that the recent push for a coordinated national AI framework should include hard rules for the tools most easily weaponised against children.
The other camp is wary of leaning too heavily on regulation of the technology, warning that watermarks can be stripped, offshore models sit beyond Australian reach, and open-source tools cannot be recalled once released. This group puts more weight on education, digital literacy and swift, consistent consequences, arguing that the durable fix is cultural: teaching young people that creating a deepfake of a peer is a serious act of abuse, not a prank, and making sure the response is predictable enough to deter it. Both camps tend to agree on one thing, which is that the status quo, where schools improvise case by case, is failing the students it is meant to protect.
What it means for Australia
For Australian families, the stakes are unusually concrete. The country has one of the world’s most active online safety regulators and has positioned itself as a global first-mover on issues such as age limits for social media, which means the way it handles school deepfakes will be watched well beyond its shores. It also means expectations are high at home. Parents who have been told the system is world-leading will not accept a shrug when their child is the one depicted in a fabricated image.
There is a fragmentation problem, too. As FluentSea has reported, Australia’s response to generative AI in education has been criticised as scattered, with different states, sectors and schools moving at different speeds on everything from teacher training to acceptable-use policies. Deepfakes expose that unevenness sharply, because a threat that travels instantly across group chats does not respect the boundaries between a Catholic school in Perth, a public high school in regional New South Wales and an independent college in Melbourne. A student in one system can be victimised by material made in another.
The economic and reputational dimension matters as well. Australia is trying to build a credible domestic AI industry and to attract data-centre and model investment, and it is doing so while asking citizens to trust the technology. Every schoolyard deepfake scandal chips away at that trust, and it hands ammunition to those who argue the sector is being allowed to grow faster than the guardrails around it.
What is next
Expect the coming months to bring more of three things. First, further prosecutions and, likely, the first genuinely testing cases involving child perpetrators, which will shape how courts and child-protection agencies calibrate their response. Second, growing pressure on the eSafety Commissioner and the federal government to spell out clearer obligations for the platforms and tools involved, potentially as part of the broader national AI rules now under discussion. Third, a scramble by schools and education departments to issue practical guidance, from reporting pathways to classroom lessons, so that staff are not left inventing a response in the middle of a crisis.
None of that will fully close the gap between how fast the technology moves and how slowly institutions adapt. But the alternative, treating each new case as a one-off, is no longer tenable when the tools sit in every student’s pocket.
Sources: GNews via OECD AI Policy Observatory.
