Meta scraped all public Facebook and Instagram posts since 2007 for AI training

Meta has admitted to scraping all public Facebook and Instagram posts made since 2007 in order to train its generative AI model, adding to the privacy contrast with Apple Intelligence.

While the admission was made during a public enquiry in Australia, the company’s statement applies globally …

Meta’s use of user posts to train AI systems

The company has previously admitted to plans to use posts and photos to train its AI systems, but has not before agreed that it has already been doing so for a great many years.

Back in June, CEO Mark Zuckerberg revealed plans to do this, as if it were a new initiative.

The announcement has raised major concerns, after he said that the company had more user data than was used to train ChatGPT – and would soon be using it to train its own AI systems. The company’s plan to use Facebook and Instagram posts and comments to train a competing chatbot raises concerns about both privacy and toxicity.

The company was legally required to offer opt-outs in both the EU and UK, though even there the legality of making it opt-out rather than opt-in is under investigation.

No such opt-out was offered in the US, Australia, or anywhere else.

An Australian investigation into Meta’s privacy practices has resulted in the company’s admission that it has been scraping public posts, photos, and comments from both Facebook and Instagram for the past 17 years.

Meta’s global privacy director Melinda Claybaugh initially denied it.

Labor senator Tony Sheldon asked whether Meta had used Australian posts from as far back as 2007 to feed its AI products, to which Ms Claybaugh responded “we have not done that”.

But when specifically pressed, she then admitted it was true for public posts.

Shoebridge: “The truth of the matter is that unless you have consciously set those posts to private since 2007, Meta has just decided that you will scrape all of the photos and all of the texts from every public post on Instagram or Facebook since 2007, unless there was a conscious decision to set them on private. That’s the reality, isn’t it?

Claybaugh: “Correct.”

Meta didn’t scrape the accounts of those aged under 18, but agreed that it included photos of children posted by parents on their own accounts.

Photo by Timothy Hales Bennett on Unsplash