Meta Utilizes Your Public Instagram and Facebook Photos to Train Its AI

Meta has announced that it uses public posts on Facebook and Instagram to train parts of its new virtual assistant, Meta AI. Private posts, which users only share with family and friends, were not used for AI training, as reported by Reuters.

Meta also refrained from using private chats from its services in AI training and took steps to filter private data from public posts, according to Meta's President for Global Affairs, Nick Clegg.

"We've tried to exclude datasets that are rich in personal information," Clegg said, adding that the majority of data Meta uses for AI training is publicly accessible. He also cited LinkedIn as an example of a website whose content Meta consciously chose not to use for privacy reasons.

Clegg noted that Meta has imposed content safety restrictions on what Meta AI can generate, such as a ban on creating photorealistic images of public figures.

Regarding copyrighted materials, Clegg mentioned expecting a fair amount of legal challenges regarding whether creative content falls under the doctrine of fair use, which allows limited use of protected works for purposes like commentary, research, and parody.

Clegg's comments come at a time when major technology companies like Meta, OpenAI, and Google have faced criticism for using internet-derived information without permission to train their artificial intelligence models.

Meta has created an assistant using its own model based on the powerful language model Llama 2, which the company released for public commercial use in July. The company also employs a new model called Emu, which generates images in response to textual prompts.

The product will be capable of generating text, audio, and images and will have real-time information access through a partnership with Microsoft's Bing search engine.

Meta AI is the most significant product among the initial consumer-oriented AI tools introduced by the company's CEO, Mark Zuckerberg, at the annual Meta Connect conference. This year, the conference focused on artificial intelligence, unlike previous conferences that emphasized augmented and virtual reality.