Three changes. One department. No legislation.

Issue AI-era guidance under the exception the policy already contains

The prohibition and its own exception sit in the same subsection of PD-2011-0418-01, under the heading “Keep the school account open”. The prohibition reads: school accounts “must not restrict access or be set as ‘private’ or ‘closed.’” Three sentences later:

A school may therefore already restrict its account to manage a risk. What no school has is a departmental statement that AI scraping and deepfake generation is such a risk. Until the department says so, a principal who restricts an account in response to the eSafety advisory is departing from a written instruction on their own judgement, and most reasonably will not.

The guidance should:

• confirm that the risks in the eSafety advisory of 28 July 2026 are risks for which restricting a school account is permitted under the existing exception;
• adopt eSafety's pre-publication checklist as departmental practice, including its two settings questions: “Is the account public?” and “Do archived pages or past posts still need to remain public?”;
• name the alternatives schools already have, including the department platforms the procedures already require for parent communication;
• address the historical archive, which stays publicly accessible even after a school stops posting.

eSafety has already said whose job this is: “schools or school sectors can set common expectations and share practical guidance, reducing the need for each school to build its response from scratch.” And: “school sectors can give principals clear guidance and shared language.”

Implementation: Departmental guidance. No amendment, no consultation, no legislation.

Amend “Keep the school account open” to permit restricted settings

The stated rationale for the clause is that “the main purpose for using a public platform is to reach a broader audience and build a stronger community, which includes extended family and friends of students and people in the local area.”

The same procedures then instruct schools not to use those accounts “as a primary method of communicating with parent or carers”, and to use department platforms instead. The open setting is therefore not what serves the school community. It adds reach to people outside it, and to every automated system that crawls the public web.

That rationale also predates all of the following:

• eSafety's advisory of 28 July 2026, and the 100-plus reports it received in a single quarter about anonymous accounts targeting schools, almost all built from imagery harvested from school social media or websites
• Meta confirming under oath that it scrapes all public posts for AI training (September 2024)
• Facial recognition companies scraping 50 billion+ photos from Facebook (confirmed by OAIC)
• 362 Australian children found in AI training datasets (Human Rights Watch, July 2024)
• Explicit deepfakes of Australian schoolgirls generated from publicly accessible photos (June 2024)

The clause should be amended so that restricted or closed settings are permitted, and are the default, for accounts that publish identifiable images of students. Victoria's education department already encourages restricted access. NSW is the outlier in requiring open accounts.

The department also has the means to act once rather than 1,780 times. School Facebook accounts are created by its own social media team, and “all school accounts on Facebook must be linked to the department's Business Manager”.

Implementation: Administrative policy update. No legislation required.

Rebuild the Permission to Publish form

eSafety asks schools two questions about consent: “Is consent active, informed and current? Does it cover school social media posts?” The current NSW form fails on every part: it never expires, it discloses none of these uses, and it bundles social media in with newsletters and the school website.

The current form offers a single binary choice: consent to all public publishing (website, newsletter, Facebook, media) or none. It makes no mention of AI training, facial recognition, data scraping, or deepfakes. The consent is indefinite with no renewal requirement.

The updated form should include:

• Granular consent by channel. Separate options for school newsletter, school website, social media, and external media.
• AI and data risk disclosure. Explicit statement that photos posted on public social media may be scraped for AI training, included in facial recognition databases, and used to generate synthetic images.
• Annual renewal. Consent should expire and require renewal each year, reflecting the rapidly changing risk landscape.
• Default to private. Social media publishing should require specific opt-in, not be bundled with general publishing consent.

Implementation: Form update. No legislation required.

Schools want to be told what the right thing now looks like

Nothing in these three asks is a criticism of schools or principals. The audit found the same pattern in 82% of them, uniformly across metro, regional and remote NSW, because that is what following the policy produces.

The current position leaves principals holding a risk they cannot lawfully reduce, and leaves the department carrying 1,780 exposures it has centrally required. eSafety notes that where the resulting material is not sexual it “may fall outside the legal criteria” for removal, so there may be no remedy after the fact. Guidance is cheaper than an incident, and the department is the only party that can issue it.