Australia is weighing how to best regulate artificial intelligence with lawmakers considering an overhaul of privacy laws following the revelations about Meta’s data collection practices. Meta, Facebook’s parent company, admitted to using public data from millions of Australian users without permission to train its AI models.
The admission includes nearly two decades’ worth of posts, photos, and status updates. During an Australian Senate inquiry, Meta’s Global Privacy Director Melinda Claybaugh confirmed the tech company’s extensive data harvesting practices and defended using “public data” from its platforms.
Labor Sen. Tony Sheldon voiced concerns over this practice, questioning Claybaugh on if the company had used posts dating back to 2007. She answered “we have not done that,” but the answer was quickly questioned by Greens Sen. David Shoebridge.
Shoebridge stated that unless posts were set to “private” since 2007, they have been scraped. Claybaugh answered that the statement was correct.
The ethics of using data from as far back as 2007 for AI training without explicit user consent was also questioned at the inquiry, especially when it comes to users who are underaged.
Meta claims it excluded data from users under the age of 18 and those who marked their posts as “private” but admitted it does not offer an opt-out option like it does for users in the EU.
The company also did not directly address whether images of children shared by adult accounts are included in its data scraping.
A recent investigation by Human Rights Watch revealed that hundreds of photos of Australian children, some as young as three years old, were included in a widely used AI training dataset.
This discovery has amplified calls in Australia for stronger legal protections to safeguard children’s data from potential misuse in AI technology.
