Aust Proposal Could Let AI Scrape Public Content

A draft proposal in Australia would grant AI companies the legal right to scrape publicly available online content for training purposes. This shift would happen if firms secure high-level agreements with major industry groups, rather than obtaining consent from individual users. The government is currently consulting on these changes, which aim to regularize practices that many experts believe are already occurring.

For everyday users, this means that photos, social media posts, and other creative works published online could be used to build AI models without explicit permission. The only way to prevent this under the proposed rules would be to restrict access to the content, such as by using passwords or paywalls. This effectively forces a trade-off: keep your content open to the public, or pay the price of privacy by limiting your audience.

Consent shifts to industry deals

The core mechanism of the proposal relies on collective bargaining. Instead of negotiating with millions of individual creators, AI companies would deal with representative bodies for authors, musicians, and other rights holders. According to Lisa Harrison, a lecturer at Flinders University, this approach is designed to catch up with the reality that tech companies are already scraping data. It provides a legal shield for actions that are currently in a gray area.

However, this structure creates a significant power imbalance. Individuals who do not belong to these major industry groups, such as casual social media users, have no direct say in the process. If you post on a platform, the terms of service and the platform’s agreements with AI firms determine your fate, not your personal preferences. This removes agency from the person who actually created the content.

Creators face income uncertainty

Professional creators are concerned that they may lose income without receiving fair compensation. Claire Pullen of the Australian Writers’ Guild notes that AI is already generating imitations of existing work, reducing demand for original content. Under the new rules, companies might gain broad access to a category of content through a single deal, leaving individual small-scale creators with no guarantee of payment. The payment structure remains undefined, with proposals ranging from direct deals to payments through a central distribution body.

There is a risk that AI companies will access vast amounts of data while the financial benefits are captured by large organizations or the companies themselves. For the millions of Australians who share content online for personal reasons, the likelihood of receiving any payment is even lower. The system prioritizes the convenience of data access for AI developers over the economic rights of individual content creators.

Risk of content geo-blocking

Another potential consequence is the loss of access to overseas creative works. If Australian law allows sweeping AI access to local content, foreign creators and publishers may respond by blocking Australian users from accessing their music, videos, and literature. This is a precedent seen in other regions where artists have fought to prevent their work from being used for AI training.

The Guardian reports that similar reforms were rejected in the UK after pressure from high-profile artists. In the US, legal battles over AI training data are ongoing, with significant settlements and lawsuits indicating the high stakes involved. While it is not yet clear if the Australian proposal would directly apply to foreign content, copyright law generally covers material distributed within the country. This could lead to a fragmented digital landscape where Australians lose access to global cultural products.