Meta has rolled out Muse Image and Muse Video, two new AI generation tools that can hoover up your public Instagram profile pictures and spit out whatever synthetic nonsense a user fancies, and the backlash was instantaneous.
What Meta Actually Launched
On Tuesday 7 July 2026, Meta unveiled Muse Image and Muse Video through its AI at Meta blog and the Meta Store. The company bills them as “image generation built for your world” — a phrase that should terrify anyone who has ever posted a holiday snap on Instagram. Muse Image handles stills. Muse Video handles motion. Both are trained on the vast corpus of visual data Meta sits on, which includes the billions of public photos uploaded to Instagram since 2010. If your profile is public, your face is in the training set. Your kids’ faces are in the training set. That drunken photo from Magaluf 2014 is in the training set.
Mark Zuckerberg, a man who has never met a privacy boundary he didn’t want to bulldoze, framed the launch as “empowering creativity.” What he didn’t mention is that the creative empowerment flows one way: from your digital life into Meta’s commercial products. The tools are available now. No opt-in. No consent. Just a quiet update to the terms of service that nobody reads and a blog post written in the anodyne corporate dialect that makes “we’re taking your shit” sound like “we’re enhancing your experience.”
The Outcry Was Immediate And Predictable
By Wednesday morning, the BBC was reporting “outcry” — a word that in journalism means “people are properly furious and we can’t ignore it anymore.” Privacy campaigners, digital rights groups, and anyone with a functioning frontal lobe pointed out the obvious: Meta has effectively granted every user a licence to deepfake every other user. Public Instagram profiles are not a consent form for AI training. They are not a model release. They are photos people shared with friends, family, and followers, not with a multinational surveillance company building a synthetic media engine.
The Information Commissioner’s Office has not yet commented, but sources suggest an investigation is inevitable. The UK’s data protection regime requires lawful basis for processing personal data. “It’s on the internet” is not a lawful basis. Neither is “we changed the T&Cs last Tuesday.” Meta’s legal team will argue legitimate interest. They always do. They will lose, eventually, after years of appeals and fines that amount to rounding errors on quarterly revenue. In the meantime, your face is generating content for someone else’s prompt: “Me as a Viking,” “Me kissing a celebrity,” “Me in porn I never consented to.”
How We Got Here: The Long Surrender
This didn’t happen overnight. It happened in stages, each one greeted with shrugs. First, Facebook claimed ownership of your photos through terms of service changes in 2012. Then Instagram, acquired for $1 billion, became a data harvester for facial recognition. Then came the “memories” feature, the algorithmic feed, the forced Reels pivot, the Threads launch that scraped your Instagram graph. Each step normalised the idea that your digital self belongs to Meta. Muse is just the latest and most brazen extraction.
The generative AI gold rush accelerated the timeline. OpenAI’s DALL-E, Midjourney, Stable Diffusion — they all trained on scraped web data. Meta watched from the sidelines, sitting on the motherlode: Instagram’s 2 billion monthly active users, billions of tagged, captioned, geolocated images. They didn’t need to scrape the open web. They had a walled garden full of labelled data. All they needed was the brass neck to use it. On 7 July 2026, they did.
What The Tools Actually Do
Muse Image takes a text prompt and generates an image. Standard fare. But it also accepts an image input — “reference image” in the jargon — and can use public Instagram photos as that reference. Type “cyberpunk version of @sarahjones_london” and if Sarah’s profile is public, Muse will generate it. Muse Video does the same for short clips. The Meta Store page shows examples: “animate your memories,” “reimagine your photos.” The language is warm. The implication is cold. Your memories are now raw material for Meta’s product.
Meta says users can set profiles to private. That is the “solution” they offer. Make your account private and the bots can’t see you. But that breaks Instagram’s core function for millions: creators, small businesses, artists, community organisers, people who use the platform as a public portfolio. They must choose between visibility and becoming training data. That is not a choice. That is a racket.
The Reaction: Anger, Resignation, And The Usual Non-Statements
Digital rights group Big Brother Watch called it “a staggering erosion of bodily autonomy in the digital age.” A spokesperson said: “Meta has decided that your face is their property. Parliament needs to act before this becomes the norm.” The Open Rights Group demanded an immediate suspension pending regulatory review. Labour’s shadow digital secretary posted a thread calling for “urgent clarification on consent mechanisms.” The Tories, as ever, muttered about innovation and investment.
Meta’s response? A statement so vapid it could have been generated by their own AI: “We’re committed to building AI responsibly and giving people control over their data.” No details. No timeline. No admission that “control” currently means “delete your account or suck it up.” Nick Clegg, the former Deputy Prime Minister turned Meta’s global affairs president, will likely appear on Radio 4’s Today programme next week to explain why this is actually good for democracy. He will use the word “ecosystem” fourteen times.
What Happens Next
The ICO will open an inquiry. The EU’s GDPR regulators will follow. Ireland’s Data Protection Commission, Meta’s lead EU regulator, will take eighteen months to issue a preliminary finding. Fines will land in 2028 or 2029. Meta will appeal. By then, Muse 3.0 will be generating feature-length films from your Instagram Stories. The only thing that stops this is legislation with teeth: a ban on training generative AI on personal data without explicit, informed, granular consent. Not “legitimate interest.” Not “publicly available.” Consent. Opt-in. Default off.
Until then, if you have a public Instagram profile, your face is in the machine. Go private. Delete the app. Or accept that Mark Zuckerberg thinks he owns your image — and he’s built the tools to prove it.