Meta admits quietly farming Australians' Facebook photos to train AI

Meta harvested nearly two decades' worth of public photos and posts from every adult Australian Facebook and Instagram user to train its AI models without consent, notification, or any opt-out option - exposing millions of people, including children, to commercial exploitation they never agreed to. The admission came during a Senate inquiry into the adoption of artificial intelligence in Australia, where Meta executives stated that while they do not scrape data from accounts of individuals under 18, any public photos shared by adults - including those featuring children - are included in the data used for AI training. Unlike users in Europe, who can opt out of having their data used for AI training due to stringent privacy laws (GDPR), Australians currently have no such option. Meta's global privacy director, Melinda Claybaugh, claimed that the absence of an opt-out mechanism in Australia is due to differing legal frameworks. The inquiry revealed that unless users actively set their posts to private, all public content shared since 2007 is subject to scraping by Meta. Claybaugh confirmed the practice, arguing that it is essential for developing effective AI tools. Politicians and privacy advocates raised alarms about the practice, describing it as a significant violation of personal privacy. Senator Tony Sheldon emphasised that many users are unaware their past posts could be used in this manner and called for stronger legal protections against such data harvesting. System 🤖 Llama Developer: Meta Platforms Country: Aus t ralia Sector: Multiple Purpose: Train AI models Technology: Machine learning Issue: Consent; Privacy/surveilla nce; Transparency

Date it happened

2024-09-01

Organisation involved

Meta Platforms

Product, system or model

Llama

This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.