Meta Has a New AI Image Tool, and I Already Used It to…

Meta has introduced Muse Image, an AI model designed for generating and editing images, now integrated into Instagram, WhatsApp, and the Meta AI app. While the tool offers creative features like composite photo generation and editing effects, it has raised significant concerns regarding the unauthorized use of user likenesses. By default, public Instagram accounts are accessible to Meta AI users, who can include a public username in prompts to create hyperrealistic AI-generated images of that person without their explicit permission.

What happened

The risk posed by the new model was demonstrated by a CNET reporter who successfully generated an AI image of a colleague by simply including their public Instagram username in a prompt. The process took less than a minute to create a deepfake of the colleague depicted as a pirate. In contrast, the model was unable to generate images for a private Instagram account, confirming that account privacy settings significantly impact the tool's reach.

Meta confirmed that users with public accounts can opt out of this feature to protect their likeness. By navigating to Instagram Settings, users can disable the "Allow people to reuse your content on Instagram and with AI features at Meta" toggle. Additionally, the Meta AI app provides a "Your likeness" setting where users can upload photos to help the system recognize them, allowing them to restrict who is authorized to use their image, ranging from only themselves to mutuals or everyone. While Meta claims its models include safeguards against illegal, abusive, or defamatory content, the ease with which these images were created highlights the ongoing challenge of preventing bad actors from bypassing such protections.