Meta disables image generation from public Instagram photos due to privacy scandal
Just three days — that's how long the Muse Image feature lasted, which allowed generating images based on public Instagram profiles. Meta promptly removed this option, acknowledging that it "did not meet user expectations" regarding privacy.
On July 7, the corporation introduced Muse Image — its first image generation model developed by Meta Superintelligence Labs. The tool was integrated into Meta AI and, in addition to standard generation from text prompts, offered a unique capability: adding a public Instagram account to the prompt so the neural network could use the user's open photos to create content.
However, by July 10, Meta published an update to the official announcement, stating the immediate disabling of this feature. The reason was widespread criticism from the community and human rights organizations.
Particular outrage was caused by the fact that the feature was enabled by default. Actress Hannah Einbinder publicly urged users to deactivate the option. The US actors' union SAG-AFTRA issued a strong statement, calling any mechanism that uses images without explicit and conspicuous consent unacceptable. The union characterized the launch as a "complete miscalculation in assessing public opinion regarding the obvious dangers and harm."
The Muse Image product itself, however, has not disappeared. The model continues to operate in Meta AI for generating and editing images from text prompts and uploaded photos. Moreover, the announcement mentions that Muse Image is already being used in over 30 new AI effects for Instagram Stories and in WhatsApp chats with Meta AI in a limited number of countries.
This incident is a vivid example of how rushing the deployment of generative AI products can lead to reputational losses. Meta clearly underestimated the audience's sensitivity to issues of personal data use, especially in the context of deepfake risks. I expect the company will now be significantly more cautious in implementing features related to user content processing and will likely introduce explicit opt-in mechanisms.