Meta has withdrawn a feature allowing Instagram photos to be used for artificial intelligence (AI) image generation just three days after its launch. This swift reversal came in response to mounting privacy concerns and fears of non-consensual digital replication, as the feature was automatically applied to public accounts without explicit user consent.
According to Meta’s official blog post on July 12, the company announced it would discontinue certain features of its Muse Image generation model, effective from July 10.
The retracted feature enabled Meta’s AI to create new images by referencing public photos from Instagram accounts tagged with @. However, the core Muse Image model and its general image generation and editing capabilities remain intact.
Meta stated that the goal was to provide an innovative creative tool while giving users control over how their public content could be used as reference material. However, it has received feedback indicating this feature fell short of expectations, and it has decided to discontinue it.
Earlier this month, on July 7, Meta unveiled Muse Image, its inaugural image generation model developed by the Meta Superintelligence Research Lab. This AI tool allows users to generate images by inputting text or uploading photos, with the added ability to directly indicate desired modifications on the output.
The controversy ignited when it was revealed that photos from adult users’ public Instagram accounts were automatically included in the feature. While private accounts and users under 18 were exempt, adult users with public profiles had to manually opt out to prevent their photos from being utilized.
The Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA) condemned the practice, stating that using users’ photos for AI generation without explicit prior consent is unacceptable. The union advised its members to disable the feature. Other entertainment agencies echoed this sentiment, arguing that any use of photos, likenesses, or creative works should require prior user consent.
Following Meta’s decision to retract the feature, SAG-AFTRA welcomed the move, stating that given the known risks of non-consensual digital replication, implementing a feature that encourages such practices is ill-advised.
This controversy has brought to light a critical question: does posting photos online implicitly grant consent for AI systems to use them in creating new images?
As generative AI services continue to expand their use of public content such as photos and videos, there’s likely to be growing pressure to establish prior consent as the standard practice, rather than relying on after-the-fact opt-out mechanisms.