Australia May Let AI Companies Train on Your Photos and Websites Unless You Say No
Leaked documents suggest the Albanese government is weighing a plan that would flip the default: your content is fair game for AI training unless you take steps to block it.
Key points
- Leaked Australian government documents reveal a proposal to adopt an "opt-out" copyright model for AI training, meaning content is available by default.
- Every publicly posted photo, small business website and personal blog could legally be used to train AI systems without the owner's permission.
- The proposal would reverse the current default under Australian copyright law, which requires permission before someone else uses your work.
- Australians would need to actively exclude their content rather than actively allow it.
- No final decision has been announced; the documents describe a proposal still under consideration.
Right now, Australian copyright law's default answer to "can an AI company use my stuff to train its systems" is no, not without permission. The Albanese government may be about to flip that to yes, unless you say otherwise.
Leaked government documents, first reported by The Guardian, outline what officials are calling an "opt-out model". AI training, the process by which companies feed vast amounts of text and images to their software so it learns to generate new content, currently sits in a legal grey zone in many countries. This proposal would resolve that grey zone firmly in favour of the AI industry. We covered the government's earlier moves in this space on 15 September, when local artists were already pushing back.
What would actually change for ordinary Australians?
Everything public on the internet becomes fair game. That family birthday album on a photo-sharing site, the product page for your bakery, the portfolio a freelance photographer spent weeks building: all of it could be scraped and used to train commercial AI systems without anyone asking first.
The burden shifts entirely to the creator. You'd need to add specific technical signals to your website or account settings to tell AI companies to stay away. Most people won't ever know they need to do this, which is largely the point from an industry perspective.
Small business owners are particularly exposed. A large corporation with a legal team can navigate an opt-out process. A sole trader with a basic website probably can't.
Who pays and who benefits?
The companies that benefit are the large AI developers, many of them American, who need enormous volumes of data to build large language models (the technology behind tools like ChatGPT) and image-generation systems. Training on Australian content costs them nothing under an opt-out regime.
Australians who created that content get nothing. No licence fee, no credit, no say.
There's a real argument that clearer rules help local AI startups too, but the windfall here flows overwhelmingly to companies already operating at scale, not to the nurse who posted health advice or the teacher who built curriculum resources online. The lawsuits Sony and Warner filed against Anthropic, which we reported on 29 August, show how high the stakes already are for creators when training data goes unregulated.
What happens next?
No law has passed. These are leaked proposal documents, not legislation. The Albanese government hasn't confirmed the plan publicly, and there'll almost certainly be a consultation period before anything is finalised.
That window matters. Submissions to government inquiries do get read, particularly when they come in volume from small business owners and creators rather than only from tech lobbyists.
One honest, doable takeaway: Before any law changes, add a robots.txt file to your website, a simple text file that tells automated programs not to copy your content, and check the privacy settings on every platform where you post photos or writing. It's not a legal guarantee, but it's the practical move right now.
Common questions
Does this affect content I have already posted?
Yes. The proposal covers the open internet broadly, meaning existing public posts would be included, not just things you publish after a law changes.
Can I stop AI companies from using my content today?
You can signal that you don't consent using a robots.txt file on your own site, or by adjusting platform privacy settings. Whether AI companies honour those signals varies, which is exactly why a legal framework matters.
