Meta admits to training its AI models with public info from Aussie users posted SINCE 2007 – NaturalNews.com

Meta Platforms has acknowledged harvesting public data from Australian Facebook and Instagram users dating back to 2007 to train its artificial intelligence models. This admission, confirmed during a Senate inquiry, reveals that the company systematically collected text and images from all public posts unless users had explicitly set them to private. The practice highlights a significant disparity in user protections, as Australian residents were not offered the opt-out mechanisms available in the European Union and the United States. The controversy underscores the urgent need for robust privacy regulations to govern how tech giants utilize personal information for AI development. Unlike European jurisdictions, which have implemented strict laws requiring user consent or providing clear opt-out pathways, Australia’s current regulatory framework left citizens exposed to extensive data mining. This legislative gap allowed Meta to prioritize the acquisition of vast datasets for creating "safer" and less biased AI tools without adequate consideration for individual privacy rights or consent. This issue is highly relevant to open data discussions as it illustrates the tension between data accessibility for technological innovation and the protection of individual autonomy. It serves as a critical case study on the ethical implications of scraping publicly available information for commercial and training purposes. The incident demonstrates that without explicit legal boundaries, the assumption that public data is freely usable for AI development can lead to widespread exploitation, emphasizing the necessity for clear policies that balance innovation with digital rights.

Source: naturalnews.com
Published on 2024-09-24