Meta Admits Scraping Public Facebook and Instagram Posts Since 2007 for AI Training

Meta Admits to Scraping Public Facebook and Instagram Posts for AI Training

• Admission and Scope: Meta has officially admitted to scraping all public posts, photos, and comments from Facebook and Instagram since 2007 for training its generative AI model. This admission was made during a public inquiry in Australia but applies globally.
• Previous Statements: The company had previously hinted at plans to use user data for AI training but never explicitly confirmed the ongoing practice. CEO Mark Zuckerberg's June announcement about using more data than ChatGPT seemed like a new initiative, but it was actually a continuation of long-standing practices.
• Privacy Concerns: This revelation has raised significant privacy concerns, especially given that Meta possesses more user data than required to train ChatGPT. The use of public posts and comments for training an AI chatbot is particularly concerning due to potential issues with privacy and toxicity.
• Legal Opt-Outs: In the EU and UK, users are legally entitled to opt-out of their data being used for AI training, though the legality of making it an opt-out rather than an opt-in system is under investigation. No such opt-outs were provided in the US, Australia, or elsewhere.
• Inclusion of Children's Photos: Meta did not scrape accounts of users under 18 but acknowledged that photos of children posted by parents on their own accounts were included.