The influencers at Kyle Germeses’ talent management company build their Instagram presences based on real-life experiences. Their control over their images and likeness became even more precarious on Tuesday when Instagram’s parent company Meta announced Muse Image, a tool that can take Instagram’s public photos and use AI to generate new images from them.
Many accounts were opted in by default, and their public photos were made available to the tool, setting off alarm bells in the entertainment industry.
“I think it’s once again wrong to expect people to literally opt out of something that has been proven to potentially cause harm,” said Hermeses, CEO of influencer talent management company G&B.
Major entertainment agency CAA said it had raised concerns with Meta on behalf of its clients.
“We call on Meta to make protections for Muse Image AI the default, not the exception, and to allow individuals to opt-in if they want to allow their images and likenesses to be used in the creation of AI content,” CAA said in a statement. “Artists have the right to decide whether and how their likeness and work will be used, with consent and the ability to set their own terms. This means creators can impose limits, monitor usage, and prevent unauthorized endorsement and abuse.”
“CAA believes in the power of new technology, but not at the expense of individual rights and livelihoods,” the agency said. “The future of creativity depends on respecting the ownership and autonomy of those who make it possible.”
SAG-AFTRA was also critical of the development.
“This type of use of Instagram users’ images is unacceptable for anything other than a clear and visible opt-in, and is a complete miscalculation of public sentiment regarding the clear danger and harm inherent in such use,” the performers’ union said in a statement.
Hollywood has long been wary of AI following a series of deepfakes, including videos and images depicting celebrities doing or saying things they didn’t approve of. Jamie Lee Curtis and others are appearing in ads for products they never endorsed. Last year, OpenAI’s video tool Sora 2 sparked outrage in Hollywood after users invoked dead celebrities without the consent of their heirs. OpenAI has since announced that it will provide rights holders with more control.
In a blog post, Meta described Muse Image as “a creative partner that knows your world inside out and turns your ideas into high-quality visuals that can be easily downloaded and shared anywhere.” In a promotional video, the company showed examples of adding friends to a photo of a band and designing furniture for your garage. Mehta said the AI-generated images are watermarked and users can report problematic images.
“We built Muse Image with strong controls and safety guardrails from day one,” Mehta said in a statement. “Private accounts and accounts owned by users under 18 are automatically excluded, and adult users with public accounts can opt out with just a few clicks. We take action against content that violates our community standards.”
Users can opt out by going to Settings and selecting “Share & Reuse” and turning off the option to allow others to create and reuse their content.
Meta said it has safeguards in place to prevent Muse Image from creating content that violates its policies, including violent, sexual, and defamatory images.
The announcement fits into the familiar Silicon Valley pattern of shipping the product first and asking for forgiveness later. This means that Meta will unlock the vast amount of content already on the platform to power new AI tools.
“They leverage their scale to not only make it easier to use their tools, but also make it easier to scale out the content that is available,” said Mickey Maher, chief business officer at Vermillio, which tracks people’s digital likenesses and intellectual property. “This is not unique to this meta product.”
Some said opting out should be the default.
“This dark pattern of overreach of essentially unfettered AI when it comes to content and information is actually something no one wants,” said Lori Fena, former president and executive director of the Electronic Frontier Foundation and co-founder of New York-based Personal Digital Space. What we need in this new AI ecosystem is the ability to create trust and have some kind of understanding and trustworthiness, which does exactly the opposite. ”
