New Muse, Old Habits: Meta pulls a feature from its AI image generation model because of privacy concerns

Meta finally made its presence felt in the AI world when it released Muse Image, an image generation model. And then prompt­ly, in keeping with its history, ran up against privacy concerns. It inte­grated Muse into Instagram, but also slipped in a feature whereby users could take public photos and videos of an account, and make new creations. The only issue was that this option was switched on by default, which meant consent was assumed unless the user deliberately opted out. Just the mention of a handle in a prompt was enough to allow someone’s content to be material for anyone.

Those affected, especially celebri­ties, didn’t take kindly to it. Agencies and unions representing actors were loud in their concerns, pub­licly criticising the feature. In India, the Ministry of Electronics and Information Technology said it would look at whether the feature was within the rules. Meta eventually did damage control and pulled the feature out. It said in a statement: “We’ve heard the feedback that this feature missed the mark, so it’s no longer available.” But its AI plans are going at speed nevertheless. It has launched Muse Spark 1.1, a thinking model that will now compete with the likes of other established players on being an agent to perform tasks.