Meta Confirms Use of Public Facebook & Instagram Posts for AI Training

Meta has recently acknowledged that it has been using publicly shared posts and photos from Facebook and Instagram for artificial intelligence training since 2007. This revelation comes amidst growing concerns over data privacy and the extent to which user content is being leveraged without explicit consent.

Data Collection from Public Posts

During a local government inquiry in Australia, Meta’s global privacy director, Melinda Claybaugh, confirmed that all publicly available posts and photos from Facebook and Instagram users have been used to train Meta’s AI models. This includes content dating back to 2007, unless users specifically set their posts to private. Claybaugh initially denied the claim but later relented after further questioning from Green Party senator David Shoebridge.

Privacy and User Consent Concerns

Meta’s practice of scraping public posts for AI training has raised questions about user consent. Many users, especially those who posted content in 2007, were unaware that their photos and posts would be used in this manner. Meta's privacy center acknowledges the use of public posts for training AI but has not clarified how long it has been collecting this data or the extent of the usage. Changing the privacy setting of posts to anything other than 'public' will stop future data collection, but it won't erase data that has already been gathered.

Limited Opt-Out Options

In regions like the EU and Brazil, privacy regulations have protected users from having their data used for AI training without consent. European users can opt out, and Brazilian users’ data cannot be used for AI purposes. However, billions of users outside of these regions remain vulnerable to having their data leveraged for Meta’s AI development without the ability to opt out.

Users in other parts of the world, including Australia and the U.S., are unable to do so if they keep their posts public. When asked if Australian users would be able to opt out in the future, Claybaugh did not provide a clear answer, citing differences in regulatory frameworks.

Concerns About Minors’ Data

There are also concerns about the potential use of data from minors. Although Meta claims it does not scrape data from users under 18, it remains unclear whether public photos of children posted by adult account holders are used for AI training. Labor Party senator Tony Sheldon raised the issue, asking if his children's public photos would be scraped, to which Claybaugh confirmed they would.

Ongoing Concerns in Australia

Australian lawmakers, including senator David Shoebridge, have expressed frustration over the lack of privacy protections for their citizens. Shoebridge pointed out that if Australia had privacy laws similar to the EU, its citizens’ data would also be safeguarded from AI training. The government’s inaction, according to Shoebridge, has left Australians exposed to continued data exploitation by companies like Meta.