When Meta announced that it would use posts on its social media platforms to train its AI, it also provided an opt-out, but then paused plans in response to public concern https://zorva.info/2024/06/21/metas-plan-on-hold-to-train-gen-ai-with-user-data/.
It seems that this did not apply to other countries, and a report from ABC News in Australia suggests that Facebook has admitted to scraping every Australian adult user’s public photos and posts to train AI, with no opt-out option unless posts were marked as private. This includes photos of children.
You can read the report, and interview with Meta’s global privacy director, at https://www.abc.net.au/news/2024-09-11/facebook-scraping-photos-data-no-opt-out/104336170.
The ICO Children’s code states that settings must always be ‘high privacy’ by default but Meta’s defence seems to be that they need lots of data to train their AI and remove biases.