Meta, the parent company of Facebook and Instagram, has been using Australian users’ posts and photos to train its artificial intelligence (AI) models since as early as 2007.
This revelation has sparked widespread concern regarding user privacy, data scraping, and the limitations of Australia’s online privacy laws.
Meta’s AI Scraping Explained
Meta’s global privacy director, Melinda Claybaugh, admitted during an inquiry held on September 10, 2024, when asked about Meta’s role in AI development and its associated privacy implications.
She confirmed that posts and photos from Australian users aged over 18 had been scraped to feed large language models (LLMs), which are integral to the development of AI technologies.
While individuals under 18 are exempt from this data collection, there’s an alarming catch. Photos of minors uploaded by parents or other adults are still subject to scraping. This has raised serious concerns about the protection of children’s images on the platform.
What’s most unsettling for many is that there is no opt-out option for Australian users. This is in stark contrast to the European Union (EU), where strict data protection laws allow users to prevent their content from being used in AI training.
In Australia, unless users manually set all posts to private, their content can be harvested by Meta for AI purposes.
The Extent of Meta’s Data Usage
Meta’s AI development practices were not a secret; the company had already disclosed in a 2023 blog post that they used publicly available content from users to train their AI models.
However, the scale and historical depth of this scraping are what caught many by surprise. The fact that posts from as far back as 2007 were included highlights a larger issue of informed consent and transparency.
Even if users now decide to make old posts private, the damage might already be done. Reports from earlier in 2024 indicated that once user data is fed into an AI model, it becomes part of the system, and making the posts private later will not remove them from the AI’s database.
Future posts marked private won’t be included in AI training, but historical posts already scraped will likely remain in the system.
Australia’s Online Privacy Laws Under Scrutiny
Australia’s lack of an opt-out feature, which is available in the EU, puts its users at a significant disadvantage. This has led to growing frustration with Australia’s outdated online privacy laws.
With an impending review of the Privacy Act, lawmakers are expected to address these growing concerns, and there’s even talk of restricting social media access to children altogether. However, details about these potential regulations, including the age limit, remain unclear.
For Australians, this revelation means that their content—posts, photos, and potentially even images of their children—has been feeding Meta’s AI algorithms for over a decade. This reality has been unsettling for many who were unaware of Meta’s extensive use of their data.
Meta’s Competitors and the Industry Trend
Meta’s practices are not unique. Other social media giants, like X (formerly Twitter) and Reddit, also allow user data to be scraped for AI training. X has offered an opt-out feature globally since July 2024, while Reddit data is being used to develop Google’s Gemini AI.
The rise of AI technologies has brought a new wave of data harvesting practices into the spotlight. These practices have fueled debates about the ethics of using user-generated content without explicit consent.
Critics argue that while companies like Meta claim they are not using private posts or direct messages, the sheer volume of publicly available data being utilized without user knowledge is problematic.
What This Means for Australian Users
For Australian users who have always kept their content set to private, there’s little to worry about. Meta has stated that private posts are not included in their AI training.
However, for those who have shared publicly available content over the years, the situation is more complicated. There’s no option to remove their content that has already been fed into AI systems.
While many users may not be concerned about casual posts being used for AI development, the broader issue lies in the lack of control over personal data.
Longtime users of Facebook and Instagram, particularly those who started using the platforms in the early days, may feel disempowered knowing that their content has been used without their explicit permission.
From a consumer rights perspective, it’s clear that Australia’s privacy laws are due for an update. The current system, which allows companies like Meta to use personal content for AI development without an opt-out option, is outdated and in need of reform.
The situation underscores the need for stronger protections and more transparent communication from tech companies regarding how user data is being used.
Will Opt-Out Options Be Introduced?
With Australia’s Privacy Act under review, an opt-out feature may be introduced in the future. The global conversation around AI and data privacy is evolving, and companies like Meta will likely face increased scrutiny over their practices.
For now, however, Australian users remain in a difficult position. Their content is still being used to train AI models, and there’s little they can do to prevent it—unless they set all future posts to private.
This lack of control has left many feeling frustrated and powerless, especially given that Meta’s rivals, like X, have already implemented opt-out options.
