Meta suddenly retracts new ‘gross’ AI feature. Is this customer-responsive or damage control?
When Instagram rolled out a new feature that allowed users to make AI “remixes” of other users’ images last week, the reaction was swift. Users didn’t like it. But the company’s walkback three days after the announcement might not have been as complete — or as user-focused — as it seemed.
At the same time Meta announced its first foray into image generation models, with Muse Image, the company announced a feature that was touted as making “sharing Stories even more fun with over 30 new effects.”
It used “advanced reasoning to understand prompts and lets you turn your ideas into high-quality, relevant visuals that you can share directly to your chat, story, or feed,” according to the product announcement on Instagram. The company suggested app users turn to the tool as they visualized “redesigning your living room.”
The remix fixture generated a lot of attention for Meta and may have also inspired some counter-strategy from Patreon, one of Instagram’s key competitors in the content creation industry.
Meta’s update also allowed users across the platform to make AI-enabled edits to any public user’s content just by tagging them.
“PSA. If u have a public Instagram, Meta just enrolled you in something gross,” a popular news account posted to its 3.3 million followers. They wanted to let anyone “create a deepfake of your face without your active knowledge or consent,” Catharina Doria posted.
“It’s literally as simple as writing a tag and a prompt, and the image likeness of that person will come up on AI,” said a presenter for the Good Law Project, a British nonprofit. “That is the world we’re living in now.”
The Good Law Project also offered video instructions on how to turn off the feature, which was turned on for all users by default.
Why would tagging be a problem?
“This automatic opt-in for the tool raises a lot of questions,” social media expert Sarah Whittle posted. “Like what if a public profile is posting photos of people who don’t consent to this? What about children’s faces? What about creators that shoot content in public without consent? In a world where mass surveillance is very much present and growing, these moments remind us that social media is not inherently a safe space.”
It wasn’t just Instagram users who pushed back. SAG-AFTRA made a statement against the tool and a recommendation “that #SagAftraMembers (and all Instagram users) opt-OUT of Meta’s new AI image generation tool, Muse Image,” on X.
“Take action to protect your likeness,” the performers’ union said. Creative Artists Agency also posted a statement against the product.
Meta removed the tagging part of the feature on July 10, three days after its launch.
“Earlier this week, we announced that one way for people to generate images in Meta AI is by @-mentioning public Instagram accounts that they want to reference,” the company said, in an update to its product announcement post. “We’ve heard the feedback that this feature missed the mark, so it’s no longer available.”
But the story didn’t end there.
“I think there’s a little bit of a misconception about what Meta pulled,” said Clara Fulks, a content creator and the founder of a digital literacy consultancy.
While Meta killed the feature that allowed public accounts to be tagged, the Muse Image feature, and the possibility of “remixing” other users’ content, is still available, and still turned on by default.
“You still do have to opt out,” Fulks told Straight Arrow.
This was the fastest Fulks had ever seen Meta pull a product feature, but the “reaction to them pulling this feature has been pretty big, considering the underlying use is still there,” she said.
That helped the company in the press, said Fulks. Now the discussion focuses on one part of the product design, rather than on the image generator itself, she said.
And, she added, Meta did not remove any content that had been made while the tagging feature was still live.
Was the retraction actually due to public pressure?
“Meta has a pretty big presence in India,” Fulks told Straight Arrow.
The Economic Times, an Indian outlet, reported last week that the Indian government planned to examine whether the image-generation tool complied with its stringent deepfake laws. India changed its deepfake laws in February to more strictly regulate what it describes as “synthetic media” and to mandate labeling and traceability of algorithmically altered content. That regulation means that anyone who generates illegal content must take it down within three hours of notification — a drastic reduction from the previous 36-hour standard.
X’s Grok chatbot had similarly been under scrutiny in the country after it generated nonconsensual sexual images of women and children.
But the response to the product and the possibility of non-consensual AI “remixing” reveals the contours of the ongoing debate around AI and social media. Meta’s terms of service still say that content posted on the platform can be used for training the company’s AI models, and Muse Image is still available to users.
Meta has not responded to Straight Arrow’s request for comment.
Who else is changing their AI-scraping policy?
While Meta was experimenting with its new feature, the CEO of Patreon took an opposite tack.
Last Thursday, Patreon CEO Jack Conte announced the company would begin blocking AI scrapers.
His instagram post, with the caption “Yaaaaasssssssss yes yes yes go go go yes yes yes get off of my cloud”, announced that “Patreon has partnered with an internet infrastructure company called Cloudflare to block AI training crawlers from using the work you publish on your Patreon to train their AI models.”
Cloudflare’s “Crawl Control” technology “allows our customers to understand AI crawler activity and then choose to allow, block or monetize that activity,” a representative for Cloudflare told Straight Arrow in an email. Cloudflare has had this capacity for a while; Crawl Control was launched in July 2025.
This could be a strategic move on the part of the platform, Fulks told Straight Arrow.
“It’s a marketing positioning issue as well, in a positive way, I think, from Patreon,” she said. “It’s kind of marking a movement, a shift in new media from platforms that value creators’ intellectual property and privacy and control over their data, versus platforms that don’t.”
Round out your reading
- With more prison space than prisoners, correctional facilities are shutting down.
- ‘Game changer’: New cancer treatment turns major surgery into an outpatient procedure.
- Viewers overwhelmingly support ‘The View’s’ fight against the FCC. Will that matter?
- Why an ICE shooting in Houston isn’t leading to mass protests.
- Milwaukee detective is the latest officer charged with misusing Flock cameras.
Ella Rae Greene, Editor In Chief
