Meta locks controversial AI photo generation feature after three days of release

"Our intention is to provide useful creative tools that give people control over how others connect to their public content. We received feedback that the feature did not meet expectations and have disabled it completely," Meta said on July 10.

The announcement comes three days after the company launched its free imaging tool Muse Image - the first imaging model from Meta Superintelligence Labs. One notable feature in Muse Image is allowing photos of other Instagram accounts to be edited, as long as the content is set to public. Users just need to tag an account, then Muse will take the photos in that account to create AI products.

Meta's policy states that "people can use your Instagram content to create new content using Meta's AI engine" and users also do not receive a notification if this action takes place.

Immediately after launch, the above feature caused privacy concerns. Meta then affirmed that if you want to prevent others from creating AI photos from Instagram posts without switching your account to private mode, users can adjust it in the settings.

Hannah Einbinder, an Emmy Award-winning actress, criticized the feature, saying it was automatically activated and urged followers to quickly disable it.

The Screen Actors Guild - American Federation of Television and Radio Artists (SAG-AFTRA), the union representing about 170,000 actors and media professionals, also urged members not to use Muse Image's new feature.

"There is no acceptable solution other than allowing Instagram users to make clear and transparent choices about how their images are used. This is a serious mistake," the organization said.

SAG-AFTRA later applauded Meta for its decision to remove the above feature from the Muse Image tool.

In addition to Muse Image, Meta also announced that it is developing Muse Video and will launch soon. The tool has the ability to comply well with commands, achieving high image fidelity and consistency over time. The development team is improving the bad parts like audio-visual synchronization and physically accurate fast movements.

Previously, Meta used third-party models like Midjourney and Black Forest Labs to power many of the photo and video creation features in Meta AI. The company said it will use its new model to reduce dependence on similar technologies.

Diep Anh (According to Reuters, Wired)