Meta has scrapped a controversial Instagram AI feature just three days after launching it, following a swift wave of criticism from privacy advocates, actors, and Hollywood's biggest talent agencies.
On July 7, Meta rolled out Muse Image, the first image-generation model from its Meta Superintelligence Labs division, built directly into the Meta AI chatbot. Among its features was a tool that let users generate images by @-mentioning any public Instagram account — effectively letting anyone reference another person's likeness in AI-generated content.
The hitch was that it was an opt-out feature rather than an opt-in one. Unless the account holder deliberately went into settings to disable it, public accounts belonging to users over the age of 18 were automatically usable. Critics claimed that although kids and private accounts were automatically excluded, the carve-out was insufficient to provide enough security.
There was immediate criticism from a number of sources:
No one's name, picture, or likeness should be used by AI without explicit, documented agreement, according to CAA, the talent agency that represents celebrities like Tom Hanks and Meryl Streep.
The actors' organisation, SAG-AFTRA, called the default settings "an utter miscalculation of public sentiment" and asked members to opt out.
Actor Hannah Einbinder (Hacks) made it known that the feature had been activated for her account automatically.
By July 10, Meta had reversed direction. A corporate spokeswoman stated that the intention was to create a helpful creative tool while providing users control over their work, but that the release "missed the mark" and that the function was no longer available.
Notably, just the Instagram-referencing feature was removed; the fundamental Muse Image text-to-image generator remains operational for ordinary prompts. Meta has not stated whether or when it plans to relaunch a version of the service with enhanced consent measures.
For more updates, join/follow our WhatsApp, Telegram and YouTube channels.