Meta has launched Muse Image, its first AI image generator built entirely by its own team. It is live inside the Meta AI app, Instagram Stories, and WhatsApp in select countries, with Facebook and Messenger to follow. For anyone who manages social media content or runs paid ads on Meta's platforms, the practical effect arrives within weeks.
Until today, when you generated an image inside a Meta app, the work was actually done by outside companies. Meta had a licensing deal with Midjourney, and separately with Black Forest Labs, whose technology it paid $140 million to use. Muse Image replaces both of those arrangements. Meta is pulling the capability in-house and cutting the payments to third parties.
This matters because of where Meta is taking it next. In the coming weeks, advertisers and agencies will be able to use Muse Image directly inside Meta's Advantage+ ad tool, which already handles automated ad creation for millions of businesses. The model can interpret a creative brief, generate image variations, adjust style, and produce brand-consistent ad versions without a designer or an outside image tool. More than 8 million advertisers currently use at least one of Meta's AI ad creative tools, so this update reaches a very large base.
For companies that run Meta ads and currently use a design agency, a freelancer, or a paid image tool to produce ad visuals, that workflow is about to face real competition from a free, built-in alternative. The output will not replace every creative brief, but for straightforward product images, background swaps, and ad variations, the case for paying separately gets harder to make.
Meta's own testing places Muse Image below OpenAI's latest image tool in overall quality, but ahead of Google's competing model on photo editing tasks. That is a second-place position, not a dominant one. The quality gap matters less than the distribution advantage: Muse Image lives inside the apps where billions of people already spend their time, and where advertisers already buy space.
The Instagram tagging feature is worth noting. Users can tag any public Instagram account by username and pull that person's public photos into a generated image. Meta says it is opt-out, meaning your company's Instagram photos could be used by anyone unless you actively go into settings and turn it off. For brands with strong visual identities, checking that setting is worth five minutes.
Behind the consumer launch is a larger shift in how Meta is building. The company spent heavily and messily over the past year, poaching researchers from OpenAI and Google with compensation packages reportedly reaching nine figures, restructuring its AI division multiple times, and spending up to $72 billion on infrastructure in 2025 alone. Muse Image is the second significant product from Meta Superintelligence Labs, the unit created to pull all of that together, following the launch of Muse Spark, its language model, in April.
A video generation model called Muse Video is already in development, with no release date given. The pattern is clear: Meta is building its own set of AI creative tools, making them free up to a point, and connecting them directly to the ad system that generates almost all of its revenue. For any business that touches Meta's advertising network, this is worth watching closely.