The parent company of Facebook and Instagram, Meta, recently admitted that it has been training its internal artificial intelligence (AI) system by using the public posts of users on both platforms without their knowledge or consent.
As reported by Fox News, the AI program, Meta AI, is ostensibly supposed to be able to respond to questions via text with its own text messages, as well as produce photorealistic images upon request. Currently available on Facebook Messenger, Instagram, and WhatsApp, the program will soon also be available on Meta’s Ray-Ban “smart glasses” and Quest 3.
Nick Clegg, Facebook’s president of global affairs, said that text and photo posts from Facebook and Instagram are being used to train Meta AI, with the posts being selected based on popularity and engagement. Clegg claimed that any personal details are removed from the posts before they are fed to the AI, with additional safeguards being added in to prevent abuse and misuse of the information provided.
However, some have raised concerns over breaches of privacy, as users did not consent to the use of their material in the training of this new AI. Others have expressed skepticism over Facebook’s claims that the AI won’t generate harmful content, such as infringement of intellectual property rights.
Facebook currently has not yet revealed how many posts have been used to train Meta AI, or how it will eventually inform users of which posts have been used for training purposes. The company has said that it will continue using public posts to train the AI, but did not say how it will handle requests from users who wish to opt out of their content being used.