Is this actually a big deal?
Common doubts, addressed with evidence. The short answer is yes.
The short version
Your school posts photos of your kids on a public Facebook Page. Those photos are visible to the entire internet. Meta confirmed under oath to the Australian Senate that it scrapes every public post to train AI. Clearview AI scraped 50 billion photos from Facebook for facial recognition. Human Rights Watch found 362 identifiable Australian children in a single AI dataset, from less than 0.0001% of the data. AI-generated explicit images of 50+ Australian schoolgirls were created from their publicly accessible photos.
None of this is theoretical. All of it is documented and sourced. The school consent form that parents signed makes no mention of any of it.
This sounds like scaremongering
A fair reaction. Here is every common objection, and what the evidence actually shows.
"Every website collects data. This is no different."
It is different in a specific and important way. When a website tracks your browsing with cookies, the data stays with that company for advertising. When a school posts a photo on a public Facebook Page, that photo is visible to the entire internet. Anyone can download it. Any company can scrape it. And confirmed evidence shows that multiple companies have.
This is not about cookies or ad tracking. It is about your child's face being downloaded, permanently stored in databases you have never heard of, and used to train AI systems you have no control over and no recourse against. A cookie expires. A photo in an AI training dataset does not.
Read Meta's Senate testimony →
"Facebook wouldn't let companies scrape photos."
Facebook is one of the companies doing it. Meta's global privacy director confirmed under oath that Meta scrapes all public posts to train its own AI models. And Meta failed to stop Clearview AI from scraping more than 50 billion photos from its platform. The Australian Information Commissioner investigated, found Clearview breached the Privacy Act, and ordered it to delete Australian data and stop collecting it. Regulators in France, Italy, the UK, and Greece imposed fines exceeding $100 million. Clearview has not demonstrated compliance with any deletion order. The database has grown since the orders were issued.
Read the Clearview AI evidence →
"Nobody is going to make a deepfake of my kid."
In June 2024, AI-generated explicit images of more than 50 schoolgirls at Bacchus Marsh Grammar in Melbourne were created using their publicly accessible photos and circulated online. In February 2025, a similar incident was reported at Gladstone Park Secondary College. Reports emerged from Sydney schools in January 2025. The eSafety Commissioner confirmed that deepfake reports are doubling year on year.
These were ordinary school photos. The kind posted on school Facebook Pages every week. The pipeline from a school Facebook post to a deepfake is short, and it is getting shorter as the tools improve and become more accessible.
Read about the deepfake incidents →
"Making it private won't change anything."
It changes specific, measurable things. A private Facebook Group removes children's photos from public web access. They stop appearing in Google search results. Third-party scrapers like Clearview AI and LAION can no longer reach them. Strangers can no longer browse or download them. Those are real protections against real threats.
Here is what it does not change: Meta still has access to content posted within its own platform, including private Groups. The only way to fully stop Meta is to get off Facebook entirely. And photos that have already been scraped from the public Page cannot be unscraped from AI models that have already trained on them.
A private Group is not a complete solution. It is the first meaningful step. And it costs nothing.
What about grandparents and family?
This is one of the first things people raise. "Nanna loves seeing the school photos on Facebook. She'll be devastated if they stop."
She won't lose access. A private Facebook Group works the same way as a public Page for anyone who is a member. Grandparents, aunts, uncles, family friends, anyone in the school community can join the Group. They request to join, an admin approves them, and from that point on they see every photo, every event, every update. It is the same experience.
The only thing that changes is that people who have no connection to the school can no longer see the content. That includes scrapers, facial recognition companies, dataset builders, and strangers.
In a public Page, only admins can post and Facebook's algorithm decides which followers see which content. Most followers never see most posts. In a private Group, all members are notified of new posts. Research shows private Groups generate higher engagement and more interaction than public Pages. Your family will likely see more school content in a Group, not less.
The school told us to use a public Page
That is true. The NSW Department of Education's social media policy, PD-2011-0418, requires school Facebook accounts to be public. The policy explicitly states accounts “must not restrict access or be set as ‘private’ or ‘closed.’” Schools did not choose this. They were directed to do it.
That policy was written in 2011. It predates Meta's confirmation that it scrapes public content for AI training. It predates Clearview AI's 50-billion-photo facial recognition database. It predates Human Rights Watch finding Australian children in AI datasets. It predates the deepfake attacks on Australian schoolgirls.
This is not the school's fault. The school followed the policy it was given. But the policy has not caught up with the technology, and the gap between what the policy assumes and what the technology does is where children are being harmed.
This is why nothing on this site asks you to argue with your school. The people who can fix it are the ones who wrote the instruction. The letter goes to your state MP and to the Minister, not to your principal.
What the department is likely to say
The objections above are the ones a parent raises. These are the ones an education department raises, and they are worth knowing before you write.
"The policy already allows schools to restrict access if there is a risk."
It does, and that is the strongest argument for the smallest ask. The clause says a school may restrict access “if required to manage or address any risks or issues”. But the sentence three lines above it says accounts “must not restrict access or be set as ‘private’ or ‘closed.’”
A principal reading both has a prohibition and an undefined exception, and no departmental statement that AI scraping and deepfake generation is one of the risks that triggers it. The audit shows how that resolves in practice: 82% of schools, uniformly across the state, keep the account open. If the exception were doing any work, the number would not be uniform.
"Public accounts build community and reach extended family."
That is the rationale the procedures give. The same procedures then instruct schools “Do not use school accounts on public platforms as a primary method of communicating with parent or carers”, and direct them to email, the school website and school apps instead.
Both cannot be load-bearing. If Facebook is not how the school communicates with its parents, the open setting is not serving the school community. And a closed group serves grandparents and family exactly as well, because they can join it.
"This is a platform problem, not a policy problem."
Partly true, and the platform problem is real and unsolved. But the department is not being asked to regulate Meta. It is being asked to stop requiring the one setting that exposes the content, on an estate of accounts its own social media team creates and links to its own Facebook Business Manager. That is entirely within its control.
"We cannot control what people copy and repost anyway."
Correct, which is why eSafety's own advice is to reduce what goes public in the first place: “posting fewer images of school students and personnel – or none at all – can reduce the risk. It cannot remove the risk completely because images may still be available from other public sources.” Nobody is claiming a closed account is a complete fix. It is the difference between a photo anyone can find and a photo someone has to work to obtain.
"There has been no incident in NSW public schools."
eSafety received more than 100 reports in the January to March 2026 quarter about anonymous accounts targeting schools and school staff, and almost all “involved imagery harvested from school social media accounts or websites”. Those accounts are usually anonymous and hosted off-platform, so a school tends to find out late. Waiting for a named local incident is a decision to be surprised by one.
What can I actually do?
It depends on how much time you have right now.
Two minutes
Enter your postcode, see how many schools in your state electorate run public Facebook Pages, and send your local member the letter. Editing even the first line so it sounds like you makes it count for more than a form letter.
Five minutes
Read the two documents side by side, then send that one link to a person who can act on it: a principal, a P&C president, a staffer, a journalist. It is a minute to read and neither document is an opinion.
At your next P&C meeting
There is a motion ready to move. It does not ask your school to change its settings. It puts the parent body on record calling on the Department of Education to change the policy, and refers the matter to the P&C Federation for state-level action. The pushback you will hear, and what to say to it, is set out with it.
This isn't about blame. It's about what we do now.
Nobody did anything wrong. Schools posted photos because parents loved seeing them, and because the policy told them to. Parents signed consent forms because they trusted the system. The system was not built for what the technology became. What is different now is that a federal regulator has written down what schools should do, and one state department still requires the opposite. That is a fixable sentence, not a fixable culture.
Follow the investigation
Get notified when new evidence emerges or policy changes.
Last reviewed: April 2026