Meta AI tool lets users generate deepfake images from public Instagram photos

Meta’s new AI feature, Muse Image, allows users to create deepfake images of others by simply entering their Instagram usernames, Danish broadcaster DR reports.

The tool, accessed through the Meta AI app, pulls from public Instagram content to generate new images. In a demonstration, a DR journalist’s public Instagram photo was used—without his explicit permission—to produce an image of him in front of a bridge, wearing a crown, and in other altered scenarios. The colleague only needed the journalist’s username, with no prior connection or formal consent required.

Christiane Vejlø, a technology analyst and podcast host, called the feature “insanely boundary-crossing,” warning it strips individuals of control over their own likeness. “Historically, we’ve had the understanding that we decide where we appear and with whom. That right is completely lost here,” she said.

Consumer advocacy group Forbrugerrådet Tænk also criticized the tool, arguing it normalizes deepfake misuse. Senior lawyer Ida Nynne Daarbak Reislev noted that Meta could have built in notifications for when someone’s images are used but chose not to. “At least then you’d have the knowledge to ask, ‘What have you done with my image?’” she said.

To limit exposure, Tænk advises users to set their Instagram and Facebook profiles to private or disable the AI feature and opt out of future content use for AI training. However, both Vejlø and Reislev condemned Meta’s default settings, which permit content use unless users actively opt out.

Meta states the feature is designed to create “creative experiences” and is rolling out across its platforms, including Instagram, WhatsApp, Facebook, and Messenger.