Meta Launches Muse Image AI Generator — and a Privacy Backlash Over Tagging Other People's Photos

Meta announced Muse Image, its first in-house AI image generator, built by Meta Superintelligence Labs. It's rolling out free across Meta AI and Instagram Stories, with WhatsApp, Facebook and advertiser tools next. The launch itself is a routine catch-up move against Google and OpenAI — but a feature that lets anyone generate AI images from another person's public photos, without notifying them, is what's actually driving the conversation.

Definition: Muse Image is Meta's new text-to-image AI model, built internally by Meta Superintelligence Labs, that also blends, edits and restyles photos.

Example: A user can @-mention someone's Instagram account inside Meta AI and generate a new image using that account's public photos — without asking first.

Key takeaway: The model itself is a normal, expected product move. The consent design of its photo-tagging feature is the actual news.

Business impact: Any brand or creator active on Instagram should assume their public photos can already be remixed by others through Muse Image, and check Meta's opt-out setting rather than assume they're excluded by default.

What did Meta announce?

Meta introduced Muse Image as the first image-generation model out of Meta Superintelligence Labs, the research group led by Alexandr Wang. The model is designed to handle complex prompts, blend multiple source photos into one image, restore and enhance old photos, transform an image's style (renaissance painting, claymation, game-asset look, and similar presets), and let users keep editing a result in the same chat instead of starting over.

It's already live inside the Meta AI app and now drives more than 30 new AI effects in Instagram Stories. Meta says WhatsApp support is live in a limited set of countries and expanding, with Facebook, Messenger, and an advertiser-facing version through Meta Advantage+ creative described as coming soon. Meta also previewed Muse Video, a video-generation model still in development, as the next step in the same product line.

What can Muse Image actually do?

Beyond basic text-to-image generation, Meta's announcement lists a specific, practical feature set: legible text rendering inside generated images, room redesigns that pull in real products from Facebook Marketplace, object and scene manipulation, and direct markup-based editing (drawing on an image to tell the model what to change). Meta also ships 30-plus preset prompts so users don't have to write one from scratch.

On pricing, Meta says Muse Image is free for "everyday creation," with higher-volume use gated behind Meta's existing paid subscription tier. Meta hasn't published the exact usage threshold where the free tier ends.

What's the photo-tagging feature, and why is it controversial?

The feature drawing the most attention lets a user @-mention any Instagram account inside Meta AI and pull that account's public photos into a new AI-generated image — including accounts that never opted into anything. Meta's own policy, as reported by TechCrunch, states that the person being tagged will not be notified that content was created using their photos.

Meta does provide a setting to turn this off, but it's opt-out: the default is on, and a person has to know the setting exists and disable it to be excluded. TechCrunch reported users pushing back immediately after launch, quoting one reaction that called it a "privacy landmine waiting to detonate."

That reaction doesn't happen in a vacuum. Meta has a specific history here: a $5 billion FTC fine in 2019 over data-handling practices, and the shutdown of Facebook's facial-recognition system in 2021 after years of scrutiny. A feature that repurposes public photos by default, with no notification, sits directly on top of that history.

How does Muse Image compare to competitors?

Meta says Muse Image performs strongly across internal benchmarks, positioning it generally ahead of Google's Nano Banana 2 and behind only OpenAI's ChatGPT image generator. That's Meta's own characterization of its own model, not an independently verified result — worth noting as a vendor claim rather than a confirmed ranking until outside benchmarks weigh in.

What should businesses and everyday users do?

Anyone with a public Instagram presence — creators, brands, or individuals — should treat their posted photos as already available for other people's Muse Image generations unless they actively opt out, since the default is inclusion, not exclusion. Checking Meta's privacy settings for this specific control is the concrete first step, not assuming a general privacy setting already covers it.

For teams evaluating Muse Image as a tool: it's free to start, already embedded in apps most consumers use daily, and functionally competitive on Meta's own account. The open question is entirely about trust and consent design, not capability — worth watching for regulatory or platform-level response before building anything that depends on the photo-tagging feature specifically.