Your Face Is in Someone Else's Prompt

Meta just shipped Muse Image with @-mention prompts that pull any public Instagram user's photos into AI creations by default — without notifying the people whose faces get used.

What's new: On July 7, Meta Superintelligence Labs launched Muse Image — its first in-house image model after years of renting Midjourney and Black Forest Labs, per CNBC. Type @username and the model composites that person's public posts and profile pictures into whatever prompt you write.

• Any public Instagram account is opted in by default to have posts, reels, and profile photos reused and AI-modified by strangers, per Business Insider.
• Instagram does not notify users when their content is tagged into someone else's AI image, per BI citing Instagram's help page.
• Opt-out requires Settings → Sharing and Reuse and toggling off reuse for posts and reels separately — settings that only appear for public accounts, per the BBC.
• Private accounts and users under 18 are excluded, a Meta spokesperson told the Straits Times.

Why it matters: Privacy advocates called it a landmine; regulators already have synthetic media in their crosshairs.

• Donald Campbell, advocacy director at tech-justice nonprofit Foxglove, told the BBC: "We've already seen a catalogue of harms from non-consensual AI-altered images on social platforms just in the past year... It is hard to see why Mark Zuckerberg thinks facilitating yet more of this creepy image manipulation is a good idea."
• Privacy International told the BBC: "Pulling real users into generated photos without explicit consent is a privacy landmine waiting to detonate."
• UK regulator Ofcom is already investigating X over Grok's role in creating and sharing non-consensual AI-altered images of real people, per the BBC.

What Meta pitched: Agentic creativity with a buried off switch.

• Alexandr Wang, who heads Meta Superintelligence Labs, said on Threads that Muse Image is "agentic" — it pairs with Muse Spark to "reason through your prompt, search the web, and plan before it generates," per The Verge.
• Meta's newsroom post pitches @-mentions for "collaborative creative concepts" and says users can turn tagging off "at any time."
• Images carry Content Seal hidden provenance watermarks in the Meta AI app, per Meta's technical blog — but nothing stops a screenshot before the tagged person knows they were in the prompt.
• Muse Image is free for everyday creation; heavier use requires Meta subscription plans. Advertisers get access through Advantage+ creative in the coming weeks, per Meta's newsroom post.

The pattern: This is Meta's second consent fight in a week — same logic, new surface.

• On June 30, WIRED reported Meta contractors on Project Cannes sent 45,000+ prompts to rival chatbots while posing as suicidal minors — without telling OpenAI, Google, or Character.AI.
• Muse Image applies the same logic to billions of public profiles: your likeness is raw material unless you find the off switch.
• Meta resumed training on EU users' public posts in April 2025 only after Ireland's Data Protection Commission paused the rollout for a year, per Reuters. Muse Image makes the extraction interactive — anyone with an @ symbol can do it.

What to watch:

• Facebook and Messenger rollout is imminent, per Meta.
• Muse Video is in preview on the same Superintelligence Labs stack.
• Expect EU and UK enforcement fast: Brussels already fined Meta €390 million in 2023 over unlawful ad-targeting data processing under GDPR, per Ireland's Data Protection Commission — and the bloc's AI Act now presses platforms on systemic risk.

Convina's view: Meta didn't invent non-consensual synthetic media — but it wired the world's largest photo archive into a one-tap @-mention prompt with opt-out buried three menus deep and zero notification. That is not a privacy glitch. It is the product. Until regulators force opt-in consent and real-time notice, every public Instagram face is fair game for someone else's imagination — and Meta's subscription upsell.