Context
Meta, the parent company of Facebook and Instagram, has found a valuable resource for training its AI models: publicly available photos and text from its own platforms.
Key Update
Meta's chief product officer, Chris Cox, revealed at Bloomberg's Tech Summit that the company utilizes public content from Instagram and Facebook to enhance its AI text-to-image generator, Emu. Cox emphasized that Meta does not use private content shared among friends, focusing only on what users have made public. This strategy gives Meta a competitive edge as Big Tech firms race to gather data to train their AI models. The Emu model, powered by this vast pool of public imagery, can generate high-quality images based on user prompts. This comes amid ongoing debates over the ethical use of data, particularly copyrighted content, with the US Copyright Office exploring new regulations to address these concerns.
